docs: July 2026 roadmap refresh

[ascerra]

Summary

Restructure the roadmap to reflect the July 1, 2026 team planning session. This replaces the June workstream structure with nine new categories prioritized by team dot-voting.

Key changes:

• At-a-glance table — visual overview at the top showing all categories with priority rank, focus area, and horizon (Now/Next/Later). Requested by the team for people who don't have time to read the full document.
• June 2026 (done) — new section documenting what shipped in June (90+ PRs merged, 150+ issues closed), replacing the stale Foundation section at the top. Covers agent registration (ADR 0058), unified env var delivery (ADR 0055), PR-based scaffold delivery, VitePress docs migration, dispatch hardening, standalone runtime progress, and more.
• Nine July categories — ordered by planning session dot-vote priority:
  1. BYOA (highest priority)
  2. Infrastructure (consolidates June's Agent Capabilities, Upgrades & Versioning, and New Forges)
  3. Observability (new — cost measurement, telemetry, hidden agent failures)
  4. Testing (behavior tests, evals, stage tests)
  5. External Partnerships (combines Docs & Public Alignment with partnership tracking)
  6. JIRA (JIRA-specific workflows, split from broader Agent Data Access)
  7. mint (extract, harden, and operationalize the mint service)
  8. Agent Data Access (data connectors, multi-repo context, credential infrastructure)
  9. Exploration (persistent agent memories, auto-merge — no votes, tracked for future)
• Foundation moved to bottom — historical context preserved but no longer dominates the top of the document
• All example bullets link to issues — no unlinked work items in "Examples of work" sections

Category mapping (June → July):

June	July
BYOA & Customizability	BYOA
Agent Capabilities & Runtime	Infrastructure (merged)
Testing, Staging Infra & Evals	Testing
Upgrades & Versioning	Infrastructure (merged)
Docs & Public Alignment	External Partnerships
New Forges	Infrastructure (merged)
Jira Integration & Refinement	JIRA + Agent Data Access (split)
(new)	Observability
(new)	mint
(new)	Exploration

Context

• Planning session: July 1, 2026 (Miro board + dot-voting)
• Companion PR for workstream-categories.md update in .fullsend repo forthcoming
• Post this PR: advertise in user forum and invite feedback per planning session decision

Test plan

• Verify all issue/PR links resolve correctly
• Confirm at-a-glance table row order matches document section order
• Review June accomplishments for accuracy and completeness
• Docs site renders correctly after merge

Made with Cursor

[qodo-code-review]

PR Summary by Qodo

Refresh roadmap structure and priorities for July 2026 planning

📝 Documentation 🕐 20-40 Minutes

AI Description

• Add an at-a-glance priority table linking to each roadmap category
• Document June 2026 shipped outcomes and reposition Foundation as historical context
• Replace June workstreams with nine July categories ordered by dot-vote priority

Diagram

graph TD
  A["docs/roadmap.md"] --> B["At-a-glance table"] --> C["June 2026 (done)"] --> D["Now: 9 priority categories"] --> E["Next: Exploration + Cross-forge"] --> F["Later: themes"] --> G["Foundation (Apr–May)"]

Loading

High-Level Assessment

The following are alternative approaches to this PR:

1. Split roadmap categories into separate files (include/links)

• ➕ Smaller diffs per category change; less merge conflict risk
• ➕ Easier ownership: CODEOWNERS per category area
• ➖ Readers lose single-scroll context unless site tooling supports includes well
• ➖ Requires additional nav/IA decisions for the docs site

2. Model the at-a-glance table as structured data (YAML/JSON) and generate markdown

• ➕ Prevents drift between table order and section order
• ➕ Enables reuse across docs site (e.g., rendered cards, filters)
• ➖ Adds a generation step and tooling/docs complexity
• ➖ Harder to edit directly in GitHub UI for quick updates

Recommendation: The PR’s single-document restructure is appropriate for a roadmap refresh and keeps consumption simple. If the team expects frequent re-ordering or many contributors editing categories in parallel, consider splitting categories into separate files next; if drift between summary/table and sections becomes an issue, move the table to generated content from structured data.

Files changed (1) +142 / -97

Documentation (1) +142 / -97

roadmap.mdRebuild roadmap around July 2026 priorities and add June accomplishments +142/-97

Rebuild roadmap around July 2026 priorities and add June accomplishments

• Reorders the roadmap to start with a new at-a-glance priority table and a June 2026 accomplishments section. Replaces the previous June workstream layout with nine July categories ordered by dot-vote priority, adds/relocates several Next/Later subsections, and moves the Foundation context to the bottom while ensuring examples link to tracked issues/PRs.

docs/roadmap.md

[github-actions]

Site preview

Preview: https://26251ece-site.fullsend-ai.workers.dev

Commit: 0f47a19376795eeb21226fcc5c8d8a734b1e0087

[qodo-code-review]

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (0) 📎 Requirement gaps (0)

Great, no issues found!

Qodo reviewed your code and found no material issues that require review

[fullsend-ai-review]

🤖 Finished Review · ✅ Success · Started 3:23 PM UTC · Completed 3:37 PM UTC
Commit: 2a3d0a5 · View workflow run →

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[fullsend-ai-review]

Review

Since the prior review (SHA b1bbd14), the branch was rebased onto main and one commit (0f47a19) restores the BYOA local-agent-runs sentence that was lost during the rebase. No substantive content changes.

All prior findings remain unresolved.

Findings

Medium

• [factual accuracy] docs/roadmap.md — The Exploration section references "ADR 0062 (#2791)" for auto-merge. ADR 0062 on disk (docs/ADRs/0062-dispatch-version-skew.md, status Accepted) is about resolving per-repo dispatch version skew — an unrelated topic. PR docs: ADR 0062 — Auto-merge #2791 is an open draft PR that proposes to supersede the existing ADR 0062 with auto-merge content. Until docs: ADR 0062 — Auto-merge #2791 merges, readers navigating to the ADR file directly (rather than following the PR link) will find content about dispatch version skew, not auto-merge. Additionally, docs/architecture.md:49 references ADR 0062 for dispatch version skew — superseding it via docs: ADR 0062 — Auto-merge #2791 would create a stale reference there. Remediation: Either assign the auto-merge ADR a new number to avoid colliding with the accepted dispatch-version-skew ADR, or clarify in the roadmap text that the ADR is pending and link only to the PR.

Low

• [broken-reference] docs/problems/platform-nativeness.md:134 — The link [Infrastructure](../roadmap.md#infrastructure) references an anchor that this same PR adds to docs/roadmap.md. Valid after merge, but fragile to future roadmap restructures that rename the section.

• [missing-authorization] docs/roadmap.md — Non-trivial change (142 additions, 98 deletions) without a linked GitHub issue. The PR body documents the July 1, 2026 planning session as the source of the restructure, which serves as the authorization record. Consider filing a tracking issue if the team wants a persistent audit trail.

---

Re-review note: All findings carry over from the prior review (SHA b1bbd14) and remain unresolved at SHA 0f47a19. The branch was rebased onto main; the only PR-specific change restores the BYOA prose sentence lost during rebase. No impact on existing findings.

Previous run

Review

Since the prior review (SHA 443e737), a single sentence was added to the BYOA prose: "Easy local agent runs let users test-drive custom agents and skills before wiring them into CI." This is consistent with the existing "Easy local agent runs" bullet in the examples list and introduces no new concerns.

All prior findings remain unresolved.

Findings

Medium

• [factual accuracy] docs/roadmap.md — The Exploration section references "ADR 0062 (#2791)" for auto-merge. ADR 0062 on disk (docs/ADRs/0062-dispatch-version-skew.md, status Accepted) is about resolving per-repo dispatch version skew — an unrelated topic. PR docs: ADR 0062 — Auto-merge #2791 is an open draft PR that proposes to supersede the existing ADR 0062 with auto-merge content. Until docs: ADR 0062 — Auto-merge #2791 merges, readers navigating to the ADR file directly (rather than following the PR link) will find content about dispatch version skew, not auto-merge. Additionally, docs/architecture.md:49 references ADR 0062 for dispatch version skew — superseding it via docs: ADR 0062 — Auto-merge #2791 would create a stale reference there. Remediation: Either assign the auto-merge ADR a new number to avoid colliding with the accepted dispatch-version-skew ADR, or clarify in the roadmap text that the ADR is pending and link only to the PR.

Low

• [broken-reference] docs/problems/platform-nativeness.md:134 — The link [Infrastructure](../roadmap.md#infrastructure) references an anchor that this same PR adds to docs/roadmap.md. Valid after merge, but fragile to future roadmap restructures that rename the section.

• [missing-authorization] docs/roadmap.md — Non-trivial change (142 additions, 98 deletions) without a linked GitHub issue. The PR body documents the July 1, 2026 planning session as the source of the restructure, which serves as the authorization record. Consider filing a tracking issue if the team wants a persistent audit trail.

Previous run

Review

Since the prior review (SHA 443e737), a single sentence was added to the BYOA prose: "Easy local agent runs let users test-drive custom agents and skills before wiring them into CI." This is consistent with the existing "Easy local agent runs" bullet in the examples list and introduces no new concerns.

All prior findings remain unresolved.

Findings

Medium

• [factual accuracy] docs/roadmap.md — The Exploration section references "ADR 0062 (#2791)" for auto-merge. ADR 0062 on disk (docs/ADRs/0062-dispatch-version-skew.md, status Accepted) is about resolving per-repo dispatch version skew — an unrelated topic. PR docs: ADR 0062 — Auto-merge #2791 is an open draft PR that proposes to supersede the existing ADR 0062 with auto-merge content. Until docs: ADR 0062 — Auto-merge #2791 merges, readers navigating to the ADR file directly (rather than following the PR link) will find content about dispatch version skew, not auto-merge. Additionally, docs/architecture.md:49 references ADR 0062 for dispatch version skew — superseding it via docs: ADR 0062 — Auto-merge #2791 would create a stale reference there. Remediation: Either assign the auto-merge ADR a new number to avoid colliding with the accepted dispatch-version-skew ADR, or clarify in the roadmap text that the ADR is pending and link only to the PR.

Low

• [broken-reference] docs/problems/platform-nativeness.md:134 — The link [Infrastructure](../roadmap.md#infrastructure) references an anchor that this same PR adds to docs/roadmap.md. Valid after merge, but fragile to future roadmap restructures that rename the section.

• [missing-authorization] docs/roadmap.md — Non-trivial change (142 additions, 98 deletions) without a linked GitHub issue. The PR body documents the July 1, 2026 planning session as the source of the restructure, which serves as the authorization record. Consider filing a tracking issue if the team wants a persistent audit trail.

---

Re-review note: All findings carry over from the prior review (SHA 443e737) and remain unresolved at SHA b1bbd14. The new commit adds a single BYOA prose sentence — no impact on existing findings.

Previous run (2)

Review

Findings

Medium

• [factual accuracy] docs/roadmap.md — The Exploration section references "ADR 0062 (#2791)" for auto-merge. ADR 0062 on disk (docs/ADRs/0062-dispatch-version-skew.md, status Accepted) is about resolving per-repo dispatch version skew — an unrelated topic. PR docs: ADR 0062 — Auto-merge #2791 is an open draft PR that proposes to supersede the existing ADR 0062 with auto-merge content. Until docs: ADR 0062 — Auto-merge #2791 merges, readers navigating to the ADR file directly (rather than following the PR link) will find content about dispatch version skew, not auto-merge. Additionally, docs/architecture.md:49 references ADR 0062 for dispatch version skew — superseding it via docs: ADR 0062 — Auto-merge #2791 would create a stale reference there. Remediation: Either assign the auto-merge ADR a new number to avoid colliding with the accepted dispatch-version-skew ADR, or clarify in the roadmap text that the ADR is pending and link only to the PR.

Low

• [broken-reference] docs/problems/platform-nativeness.md:134 — The link [Infrastructure](../roadmap.md#infrastructure) references an anchor that this same PR adds to docs/roadmap.md. Valid after merge, but fragile to future roadmap restructures that rename the section.

• [missing-authorization] docs/roadmap.md — Non-trivial change (142 additions, 98 deletions) without a linked GitHub issue. The PR body documents the July 1, 2026 planning session as the source of the restructure, which serves as the authorization record. Consider filing a tracking issue if the team wants a persistent audit trail.

Previous run (3)

Review

Both prior-review medium findings are resolved: PRs #2768/#2769/#2770 no longer appear in the BYOA examples list (internal consistency fixed), and the "Phase 4" reference in platform-nativeness.md is replaced with specific section links (stale reference fixed).

Findings

Medium

• [factual accuracy] docs/roadmap.md — The Exploration section references "ADR 0062 (#2791)" for auto-merge. ADR 0062 on disk (docs/ADRs/0062-dispatch-version-skew.md, status Accepted) is about resolving per-repo dispatch version skew — an unrelated topic. PR docs: ADR 0062 — Auto-merge #2791 is an open PR titled "docs: ADR 0062 — Auto-merge" that proposes to supersede the existing ADR 0062 with auto-merge content. Until docs: ADR 0062 — Auto-merge #2791 merges, readers navigating to the ADR file directly (rather than following the PR link) will find content about dispatch version skew, not auto-merge. Remediation: Clarify that the auto-merge ADR is pending PR docs: ADR 0062 — Auto-merge #2791, or confirm the intended ADR number once docs: ADR 0062 — Auto-merge #2791 lands.

Low

• [broken-reference] docs/problems/platform-nativeness.md:134 — The link [Infrastructure](../roadmap.md#infrastructure) references an anchor that this same PR adds to docs/roadmap.md. Valid after merge, but fragile to future roadmap restructures that rename the section.

• [missing-authorization] docs/roadmap.md — Non-trivial change (141 additions, 97 deletions) without a linked GitHub issue. The PR body documents the July 1, 2026 planning session as the source of the restructure, which serves as the authorization record. Consider filing a tracking issue if the team wants a persistent audit trail of the planning session outcomes.

• [scope-alignment] docs/problems/platform-nativeness.md:134 — The original "Phase 4 (domain specificity)" concept is mapped to the "Infrastructure" and "Cross-forge orchestration" sections. Verify this mapping preserves the original intent — domain specificity may have been broader than forge portability alone.

Previous run (4)

Review

Verdict: comment — two medium findings carried over from the prior review (both unresolved). No blockers for a docs-only roadmap restructure, but worth addressing before merge.

The restructure is well-organized: the at-a-glance table ordering matches the document section ordering, all internal links to problem docs and ADRs resolve correctly, the PR title follows the docs: Conventional Commits format, and the content aligns with the project's documented architectural principles in AGENTS.md. No security concerns, no scope creep, and no new stale references introduced beyond the two carried-forward findings below.

Findings

Medium

• [internal consistency] docs/roadmap.md — PRs #2768, #2769, and #2770 are listed as delivered work in the "June 2026 (done)" section ("Agent registration and BYOA foundations — ADR 0058 landed in three phases") and simultaneously as forward-looking work under the BYOA "Examples of work" section ("Agent registration schema and CLI subcommand"). Listing completed PRs as examples of work that "could move this forward" implies they are not yet done, contradicting the June summary. Remediation: Remove the three PRs from the BYOA examples list, or reframe them as foundational work that landed in June and will be extended — e.g., link to follow-on issues instead of the delivered PRs.

• [stale reference] docs/problems/platform-nativeness.md:134 — References "Phase 4 (domain specificity in the roadmap)" and links to ../roadmap.md. The roadmap uses Now/Next/Later structure and has never contained phase-numbered terminology. This PR's restructuring makes the disconnect more pronounced — a reader following the link will find no matching section or concept. Remediation: Update the reference to point to the appropriate roadmap section (e.g., the Infrastructure or Later sections covering forge portability), or remove the phase terminology and state the deferral directly.

---

Re-review note: Both findings were identified in the prior review (SHA 2a3d0a5) and remain unresolved in the current revision (SHA 925fadf). Severity ratings are anchored to the prior assessment.

---

Labels: PR modifies only docs/roadmap.md — a user-facing documentation restructure.

Previous run (5)

Review

Findings

Medium

• [internal consistency] docs/roadmap.md — PRs feat(config): add agent registration schema (ADR 0058 Phase 1) #2768, feat(run): add runtime agent resolution from config (ADR 0058 Phase 3) #2769, and feat(cli): add fullsend agent subcommand (ADR 0058 Phase 2) #2770 are listed as delivered work in the "June 2026 (done)" section ("Agent registration and BYOA foundations — ADR 0058 landed in three phases") and simultaneously as forward-looking work under the "BYOA" Now section ("Agent registration schema and CLI subcommand"). Listing completed PRs as examples of work that "could move this forward" implies they are not yet done, contradicting the June summary. Consider removing the three PRs from the BYOA section or rewording to clarify these are foundational work that landed in June and will be extended.

• [stale reference] docs/problems/platform-nativeness.md:134 — References "Phase 4 (domain specificity in the roadmap)" and links to ../roadmap.md. The restructured roadmap removes all phase-based terminology and has no "Phase 4" or "domain specificity" section. Readers following this link will find no matching content. Update the reference to point to the appropriate roadmap section or remove the phase terminology.

---

Labels: PR modifies only docs/roadmap.md — a user-facing documentation restructure.

Previous run (6)

Review

Findings

Medium

• [factual accuracy] docs/roadmap.md — The Exploration section references "ADR 0062 (#2791)" for auto-merge. ADR 0062 on disk (docs/ADRs/0062-dispatch-version-skew.md, status Accepted) is about resolving per-repo dispatch version skew — an unrelated topic. PR docs: ADR 0062 — Auto-merge #2791 is an open draft PR that proposes to supersede the existing ADR 0062 with auto-merge content. Until docs: ADR 0062 — Auto-merge #2791 merges, readers navigating to the ADR file directly (rather than following the PR link) will find content about dispatch version skew, not auto-merge. Additionally, docs/architecture.md:49 references ADR 0062 for dispatch version skew — superseding it via docs: ADR 0062 — Auto-merge #2791 would create a stale reference there. Remediation: Either assign the auto-merge ADR a new number to avoid colliding with the accepted dispatch-version-skew ADR, or clarify in the roadmap text that the ADR is pending and link only to the PR.

Low

• [broken-reference] docs/problems/platform-nativeness.md:134 — The link [Infrastructure](../roadmap.md#infrastructure) references an anchor that this same PR adds to docs/roadmap.md. Valid after merge, but fragile to future roadmap restructures that rename the section.

• [missing-authorization] docs/roadmap.md — Non-trivial change (142 additions, 98 deletions) without a linked GitHub issue. The PR body documents the July 1, 2026 planning session as the source of the restructure, which serves as the authorization record. Consider filing a tracking issue if the team wants a persistent audit trail.

---

Re-review note: All findings carry over from the prior review (SHA 307cb63) and remain unresolved at SHA 443e737. The factual-accuracy finding is enhanced with new evidence about docs/architecture.md:49 also referencing ADR 0062 for dispatch version skew. Severity ratings are anchored to the prior assessment.

Previous run (7)

Review

Both prior-review medium findings are resolved: PRs #2768/#2769/#2770 no longer appear in the BYOA examples list (internal consistency fixed), and the "Phase 4" reference in platform-nativeness.md is replaced with specific section links (stale reference fixed).

Findings

Medium

• [factual accuracy] docs/roadmap.md — The Exploration section references "ADR 0062 (#2791)" for auto-merge. ADR 0062 on disk (docs/ADRs/0062-dispatch-version-skew.md, status Accepted) is about resolving per-repo dispatch version skew — an unrelated topic. PR docs: ADR 0062 — Auto-merge #2791 is an open PR titled "docs: ADR 0062 — Auto-merge" that proposes to supersede the existing ADR 0062 with auto-merge content. Until docs: ADR 0062 — Auto-merge #2791 merges, readers navigating to the ADR file directly (rather than following the PR link) will find content about dispatch version skew, not auto-merge. Remediation: Clarify that the auto-merge ADR is pending PR docs: ADR 0062 — Auto-merge #2791, or confirm the intended ADR number once docs: ADR 0062 — Auto-merge #2791 lands.

Low

• [broken-reference] docs/problems/platform-nativeness.md:134 — The link [Infrastructure](../roadmap.md#infrastructure) references an anchor that this same PR adds to docs/roadmap.md. Valid after merge, but fragile to future roadmap restructures that rename the section.

• [missing-authorization] docs/roadmap.md — Non-trivial change (141 additions, 97 deletions) without a linked GitHub issue. The PR body documents the July 1, 2026 planning session as the source of the restructure, which serves as the authorization record. Consider filing a tracking issue if the team wants a persistent audit trail of the planning session outcomes.

• [scope-alignment] docs/problems/platform-nativeness.md:134 — The original "Phase 4 (domain specificity)" concept is mapped to the "Infrastructure" and "Cross-forge orchestration" sections. Verify this mapping preserves the original intent — domain specificity may have been broader than forge portability alone.

Previous run (8)

Review

Verdict: comment — two medium findings carried over from the prior review (both unresolved). No blockers for a docs-only roadmap restructure, but worth addressing before merge.

The restructure is well-organized: the at-a-glance table ordering matches the document section ordering, all internal links to problem docs and ADRs resolve correctly, the PR title follows the docs: Conventional Commits format, and the content aligns with the project's documented architectural principles in AGENTS.md. No security concerns, no scope creep, and no new stale references introduced beyond the two carried-forward findings below.

Findings

Medium

• [internal consistency] docs/roadmap.md — PRs #2768, #2769, and #2770 are listed as delivered work in the "June 2026 (done)" section ("Agent registration and BYOA foundations — ADR 0058 landed in three phases") and simultaneously as forward-looking work under the BYOA "Examples of work" section ("Agent registration schema and CLI subcommand"). Listing completed PRs as examples of work that "could move this forward" implies they are not yet done, contradicting the June summary. Remediation: Remove the three PRs from the BYOA examples list, or reframe them as foundational work that landed in June and will be extended — e.g., link to follow-on issues instead of the delivered PRs.

• [stale reference] docs/problems/platform-nativeness.md:134 — References "Phase 4 (domain specificity in the roadmap)" and links to ../roadmap.md. The roadmap uses Now/Next/Later structure and has never contained phase-numbered terminology. This PR's restructuring makes the disconnect more pronounced — a reader following the link will find no matching section or concept. Remediation: Update the reference to point to the appropriate roadmap section (e.g., the Infrastructure or Later sections covering forge portability), or remove the phase terminology and state the deferral directly.

---

Re-review note: Both findings were identified in the prior review (SHA 2a3d0a5) and remain unresolved in the current revision (SHA 925fadf). Severity ratings are anchored to the prior assessment.

---

Labels: PR modifies only docs/roadmap.md — a user-facing documentation restructure.

Previous run (9)

Review

Findings

Medium

• [internal consistency] docs/roadmap.md — PRs feat(config): add agent registration schema (ADR 0058 Phase 1) #2768, feat(run): add runtime agent resolution from config (ADR 0058 Phase 3) #2769, and feat(cli): add fullsend agent subcommand (ADR 0058 Phase 2) #2770 are listed as delivered work in the "June 2026 (done)" section ("Agent registration and BYOA foundations — ADR 0058 landed in three phases") and simultaneously as forward-looking work under the "BYOA" Now section ("Agent registration schema and CLI subcommand"). Listing completed PRs as examples of work that "could move this forward" implies they are not yet done, contradicting the June summary. Consider removing the three PRs from the BYOA section or rewording to clarify these are foundational work that landed in June and will be extended.

• [stale reference] docs/problems/platform-nativeness.md:134 — References "Phase 4 (domain specificity in the roadmap)" and links to ../roadmap.md. The restructured roadmap removes all phase-based terminology and has no "Phase 4" or "domain specificity" section. Readers following this link will find no matching content. Update the reference to point to the appropriate roadmap section or remove the phase terminology.

---

Labels: PR modifies only docs/roadmap.md — a user-facing documentation restructure.

Previous run (10)

Review

Both prior-review medium findings are resolved: PRs #2768/#2769/#2770 no longer appear in the BYOA examples list (internal consistency fixed), and the "Phase 4" reference in platform-nativeness.md is replaced with specific section links (stale reference fixed).

Findings

Medium

• [factual accuracy] docs/roadmap.md — The Exploration section references "ADR 0062 (#2791)" for auto-merge. ADR 0062 on disk (docs/ADRs/0062-dispatch-version-skew.md, status Accepted) is about resolving per-repo dispatch version skew — an unrelated topic. PR docs: ADR 0062 — Auto-merge #2791 is an open PR titled "docs: ADR 0062 — Auto-merge" that proposes to supersede the existing ADR 0062 with auto-merge content. Until docs: ADR 0062 — Auto-merge #2791 merges, readers navigating to the ADR file directly (rather than following the PR link) will find content about dispatch version skew, not auto-merge. Remediation: Clarify that the auto-merge ADR is pending PR docs: ADR 0062 — Auto-merge #2791, or confirm the intended ADR number once docs: ADR 0062 — Auto-merge #2791 lands.

Low

• [broken-reference] docs/problems/platform-nativeness.md:134 — The link [Infrastructure](../roadmap.md#infrastructure) references an anchor that this same PR adds to docs/roadmap.md. Valid after merge, but fragile to future roadmap restructures that rename the section.

• [missing-authorization] docs/roadmap.md — Non-trivial change (141 additions, 97 deletions) without a linked GitHub issue. The PR body documents the July 1, 2026 planning session as the source of the restructure, which serves as the authorization record. Consider filing a tracking issue if the team wants a persistent audit trail of the planning session outcomes.

• [scope-alignment] docs/problems/platform-nativeness.md:134 — The original "Phase 4 (domain specificity)" concept is mapped to the "Infrastructure" and "Cross-forge orchestration" sections. Verify this mapping preserves the original intent — domain specificity may have been broader than forge portability alone.

Previous run (11)

Review

Verdict: comment — two medium findings carried over from the prior review (both unresolved). No blockers for a docs-only roadmap restructure, but worth addressing before merge.

The restructure is well-organized: the at-a-glance table ordering matches the document section ordering, all internal links to problem docs and ADRs resolve correctly, the PR title follows the docs: Conventional Commits format, and the content aligns with the project's documented architectural principles in AGENTS.md. No security concerns, no scope creep, and no new stale references introduced beyond the two carried-forward findings below.

Findings

Medium

• [internal consistency] docs/roadmap.md — PRs #2768, #2769, and #2770 are listed as delivered work in the "June 2026 (done)" section ("Agent registration and BYOA foundations — ADR 0058 landed in three phases") and simultaneously as forward-looking work under the BYOA "Examples of work" section ("Agent registration schema and CLI subcommand"). Listing completed PRs as examples of work that "could move this forward" implies they are not yet done, contradicting the June summary. Remediation: Remove the three PRs from the BYOA examples list, or reframe them as foundational work that landed in June and will be extended — e.g., link to follow-on issues instead of the delivered PRs.

• [stale reference] docs/problems/platform-nativeness.md:134 — References "Phase 4 (domain specificity in the roadmap)" and links to ../roadmap.md. The roadmap uses Now/Next/Later structure and has never contained phase-numbered terminology. This PR's restructuring makes the disconnect more pronounced — a reader following the link will find no matching section or concept. Remediation: Update the reference to point to the appropriate roadmap section (e.g., the Infrastructure or Later sections covering forge portability), or remove the phase terminology and state the deferral directly.

---

Re-review note: Both findings were identified in the prior review (SHA 2a3d0a5) and remain unresolved in the current revision (SHA 925fadf). Severity ratings are anchored to the prior assessment.

---

Labels: PR modifies only docs/roadmap.md — a user-facing documentation restructure.

Previous run (12)

Review

Findings

Medium

• [internal consistency] docs/roadmap.md — PRs feat(config): add agent registration schema (ADR 0058 Phase 1) #2768, feat(run): add runtime agent resolution from config (ADR 0058 Phase 3) #2769, and feat(cli): add fullsend agent subcommand (ADR 0058 Phase 2) #2770 are listed as delivered work in the "June 2026 (done)" section ("Agent registration and BYOA foundations — ADR 0058 landed in three phases") and simultaneously as forward-looking work under the "BYOA" Now section ("Agent registration schema and CLI subcommand"). Listing completed PRs as examples of work that "could move this forward" implies they are not yet done, contradicting the June summary. Consider removing the three PRs from the BYOA section or rewording to clarify these are foundational work that landed in June and will be extended.

• [stale reference] docs/problems/platform-nativeness.md:134 — References "Phase 4 (domain specificity in the roadmap)" and links to ../roadmap.md. The restructured roadmap removes all phase-based terminology and has no "Phase 4" or "domain specificity" section. Readers following this link will find no matching content. Update the reference to point to the appropriate roadmap section or remove the phase terminology.

---

Labels: PR modifies only docs/roadmap.md — a user-facing documentation restructure.

[dhshah13]

Is GitLab integration still a priority? Or is it going to be in next month?

[ascerra]

Is GitLab integration still a priority? Or is it going to be in next month?

It is currently scoped as "next" here

So yes, its not being deferred to "later" but we did identify a lot of work that takes priority over it. Next month is a good estimate IMO.

[ggallen]

Is GitLab integration still a priority? Or is it going to be in next month?

@dhshah13, the ADR is up for review in #2583. Please review it and either add comments or approve.

As soon as this is merged I plan to start implementing around my other tasks, so I am hoping to make good progress on it this month. You approval would help get the ADR completed.

[dhshah13]

@ggallen I'll finish the review today and post my comments.

[fullsend-ai-review]

🤖 Finished Review · ✅ Success · Started 11:32 AM UTC · Completed 11:39 AM UTC
Commit: 925fadf · View workflow run →

[ascerra]

Addressed both medium findings in 307cb631:

[internal consistency] docs/roadmap.md — duplicate BYOA PR references
Removed PRs #2768, #2769, and #2770 from the BYOA "Examples of work" list. They remain documented as delivered work in the June 2026 section, so the forward-looking examples no longer imply unfinished agent-registration work.

[stale reference] docs/problems/platform-nativeness.md:134 — Phase 4 roadmap link
Replaced the "Phase 4 (domain specificity in the roadmap)" wording with links to the Infrastructure and Cross-forge orchestration sections, which now cover forge portability in the Now/Next/Later structure.

[fullsend-ai-review]

🤖 Finished Review · ✅ Success · Started 11:50 AM UTC · Completed 12:07 PM UTC
Commit: 307cb63 · View workflow run →

[fullsend-ai-review]

🤖 Finished Review · ✅ Success · Started 9:41 PM UTC · Completed 9:57 PM UTC
Commit: 443e737 · View workflow run →

[fullsend-ai-review]

🤖 Finished Review · ✅ Success · Started 2:21 PM UTC · Completed 2:27 PM UTC
Commit: b1bbd14 · View workflow run →

[Roming22]

lgtm

[Roming22]

This PR is not the right venue to discuss here, but I wanted to raise the topic that testing should be higher in the priority list. It is our opinion that we cannot start using FullSend in any significant manner unless stability is greatly improved. Priorities might be a good topic for the next user forum once you have all the pilot teams goals regarding adoption.

[fullsend-ai-review]

🤖 Finished Review · ✅ Success · Started 1:27 PM UTC · Completed 1:37 PM UTC
Commit: 0f47a19 · View workflow run →

[fullsend-ai-review]

See the review comment for full details.

[fullsend-ai-retro]

🤖 Finished Retro · ✅ Success · Started 1:49 PM UTC · Completed 1:58 PM UTC
Commit: 0f47a19 · View workflow run →

[fullsend-ai-retro]

PR #2850 was a human-authored docs-only roadmap restructure (July 2026 planning refresh) that took 8 days from open to merge. The review agent ran 6 times and found real issues — internal consistency (duplicate BYOA PR references), stale references ('Phase 4' terminology), and factual accuracy (ADR 0062 conflict) — two of which the author fixed. Human reviewers added substantive product-level feedback (timeline granularity, missing local-agent-runs content) that the agent could not provide. Three known patterns recurred: (1) the sticky review comment accumulated 29K chars with 13 nested 'Previous run' sections from 6 runs due to quadratic SHA duplication (evidence for #2358); (2) the medium-severity ADR 0062 factual-accuracy finding was merged unresolved without explicit dismissal (evidence for #2940); (3) identical findings were repeated across 4 consecutive reviews on unchanged code (evidence for #1013). All improvement opportunities map to existing tracked issues.

Proposals filed

• Evidence for #2358: sticky review comment history nesting causes quadratic duplication on PR #2850 (in fullsend-ai/fullsend)
• Evidence for #2940: medium-severity ADR 0062 finding merged unresolved on PR #2850 (in fullsend-ai/fullsend)
• Evidence for #1013: 100% finding repetition rate across 4 reviews on unchanged code (PR #2850) (in fullsend-ai/fullsend)