[automation] Update Convex AI files

[openclaw-barnacle]

Summary

• refresh Convex-managed AI guidance files
• keep AGENTS.md / CLAUDE.md Convex sections in sync when Convex updates them
• update repo-local Convex developer skills under .agents/skills

Validation

• $(bun pm bin)/convex ai-files status

[clawsweeper]

🦞👀
ClawSweeper picked this up.

Pull request received. I will update this pull request when review starts.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
clawhub	Ready	Preview	Aug 17, 2026 7:26am

[clawsweeper]

Codex review: needs changes before merge. Reviewed August 17, 2026, 3:28 AM ET / 07:28 UTC.

ClawSweeper review

What this changes

The PR refreshes generated Convex agent guidance and its recorded upstream skill revision.

Regression provenance

Possible regression — suspected (reviewed change). No predecessor PR is attributed.

Merge readiness

⛔ Blocked by patch quality or review findings - 6 items remain

Keep open: the updated head still removes the direct confirmation gate before tiered spending procedures, a P1 security-boundary regression from current main.

Priority: P1
Reviewed head: 7f54352d93e1d4bdd2e9799a22b6a5fde3a4f1ac

Review scores

Measure	Result	What it means
Overall readiness	🧂 unranked krab (1/6)	PR readiness rating was derived from proof quality, review findings, security review, and reviewer confidence.
Proof confidence	🌊 off-meta tidepool	Not applicable: Real behavior proof is not required for maintainer- or bot-authored pull requests.
Patch quality	🧂 unranked krab (1/6)	Security review found an item that needs attention.

Verification

Check	Result	Evidence
Real behavior	Not applicable	Not applicable: Real behavior proof is not required for maintainer- or bot-authored pull requests.
Evidence reviewed	5 items	Direct safety gate removed: The reviewed diff removes the tier-aware confirmation branch and rule from the directly invocable add-capability guidance. Current-main contract: Current main requires explicit confirmation before a matched tier>0 procedure and repeats the rule in the direct skill. Shared safety invariant: The root Convex guidance routes capability additions to convex-add and says tier>0 capabilities always require explicit user confirmation.
Findings	1 actionable finding	[P1] Restore confirmation before tiered spending actions
Security	Needs attention	Direct tiered-spend path loses confirmation: A matched tier>0 capability is followed without first obtaining explicit user confirmation, weakening the existing paid-action boundary.

How this fits together

ClawHub ships local Convex guidance that agents read when adding backend capabilities. A direct add-capability request selects a hosted procedure or local fallback, which can lead to paid cloud actions.

flowchart LR
  A[Capability request] --> B[Convex add guidance]
  B --> C[Hosted catalog match]
  C --> D{Tiered spending action?}
  D --> E[Ask for explicit confirmation]
  E --> F[Follow procedure]
  F --> G[Capability configured]

Loading

Before merge

• Restore confirmation before tiered spending actions (P1) - The direct convex-add path now follows every matched hosted procedure immediately. A tier>0 match can perform a paid action, but the PR removes the current-main confirmation branch and rule; the root convex skill cannot enforce that requirement when this skill is selected directly.
• Resolve security concern: Direct tiered-spend path loses confirmation - A matched tier>0 capability is followed without first obtaining explicit user confirmation, weakening the existing paid-action boundary.
• Resolve merge risk (P1) - Merging would let a directly selected tier>0 hosted capability proceed without an explicit yes; the root router’s rule is not guaranteed to apply to direct convex-add invocation.
• Complete next step (P2) - A narrow, source-proven repair can restore the direct confirmation guard and regenerate the managed files.
• Improve patch quality - Resolve the security review concern or explain why the changed path is safe.
• Improve patch quality - Address the highest-priority review finding and re-run the changed-surface validation.

Findings

• [P1] Restore confirmation before tiered spending actions — .agents/skills/convex-add/SKILL.md:16
• [high] Direct tiered-spend path loses confirmation — .agents/skills/convex-add/SKILL.md:16

Agent review details

Security

Needs attention: The diff removes a direct user-consent boundary before agent-directed tiered spending actions.

Review metrics

Metric	Value	Why it matters
Guidance refresh scope	10 files affected; +28/-48 lines	A small generated refresh contains a removal of the spending-consent guard, so the safety boundary needs focused review.

Merge-risk options

Maintainer options:

1. Restore the direct confirmation gate (recommended)
   Correct the generating Convex skill source and regenerate this PR so a tier>0 catalog match asks for explicit confirmation before its procedure runs.
2. Accept a changed spending policy
   Only merge after maintainers explicitly approve removing the direct confirmation requirement and document an equivalent enforced boundary.

Copy recommended automerge instruction

@clawsweeper automerge

Special instructions:
Restore tier-aware explicit confirmation in the generated convex-add workflow, regenerate the managed files, and add focused proof that a matched tier>0 capability pauses before following its procedure.

Technical review

Best possible solution:

Regenerate from a Convex source that retains tier-aware confirmation in convex-add, then show a redacted run where a matched tier>0 request pauses before its procedure is followed.

Do we have a high-confidence way to reproduce the issue?

Yes: inspect a tier>0 catalog match through the direct convex-add workflow; the PR head unconditionally follows its procedure, while current main pauses for explicit confirmation.

Is this the best way to solve the issue?

No. The root skill retains the policy, but direct selection routes into convex-add, so the generated source must preserve its own tier-aware gate.

Full review comments:

• [P1] Restore confirmation before tiered spending actions — .agents/skills/convex-add/SKILL.md:16
  The direct convex-add path now follows every matched hosted procedure immediately. A tier>0 match can perform a paid action, but the PR removes the current-main confirmation branch and rule; the root convex skill cannot enforce that requirement when this skill is selected directly.
  Confidence: 0.99

Overall correctness: patch is incorrect
Overall confidence: 0.99

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning high; reviewed against 4117154ecac4.

Labels

Label justifications:

• P1: The proposed direct workflow can reach tiered spending actions without the explicit confirmation required on current main.
• merge-risk: 🚨 security-boundary: The patch removes an approval boundary before agent-directed paid actions.
• rating: 🧂 unranked krab: Overall readiness is 🧂 unranked krab; proof is 🌊 off-meta tidepool and patch quality is 🧂 unranked krab.
• status: ⏳ waiting on author: ClawSweeper has contributor-facing work open and is waiting for author action. Not applicable: Real behavior proof is not required for maintainer- or bot-authored pull requests.

Evidence

Security concerns:

• [high] Direct tiered-spend path loses confirmation — .agents/skills/convex-add/SKILL.md:16
  A matched tier>0 capability is followed without first obtaining explicit user confirmation, weakening the existing paid-action boundary.
  Confidence: 0.99

Acceptance criteria:

• [P1] Run $(bun pm bin)/convex ai-files status.
• [P1] Provide redacted terminal or live output showing a matched tier>0 catalog capability pauses for explicit confirmation before its procedure is followed.
• [P1] Run the focused generation or validation command supplied by the authoritative Convex AI-files source, if available.

What I checked:

• Direct safety gate removed: The reviewed diff removes the tier-aware confirmation branch and rule from the directly invocable add-capability guidance. (.agents/skills/convex-add/SKILL.md:16, 7f54352d93e1)
• Current-main contract: Current main requires explicit confirmation before a matched tier>0 procedure and repeats the rule in the direct skill. (.agents/skills/convex-add/SKILL.md:16, 4117154ecac4)
• Shared safety invariant: The root Convex guidance routes capability additions to convex-add and says tier>0 capabilities always require explicit user confirmation. (.agents/skills/convex/SKILL.md:25, 4117154ecac4)
• Feature history: Available history identifies a prior bot-managed AI-files refresh and a later current-main touch on this direct-skill path. (.agents/skills/convex-add/SKILL.md:10, 0b6017548c30)
• Not implemented on main: The proposed head is not an ancestor of current main and no release tag contains it. (7f54352d93e1)

Likely related people:

• Patrick Erichsen: The latest available current-main history entry touching the direct capability guidance is 0b60175. (role: recent area contributor; confidence: medium; commits: 0b6017548c30; files: .agents/skills/convex-add/SKILL.md)
• openclaw-barnacle[bot]: Available history attributes the earlier managed Convex AI-files refresh to this bot. (role: introduced prior generated guidance refresh; confidence: medium; commits: 9009eae00355; files: .agents/skills/convex-add/SKILL.md, .agents/skills/convex/SKILL.md)

Rating scale

Score	Internal tier	Crab rank	Meaning
6/6	S	🦀 challenger crab	Exceptional readiness
5/6	A	🦞 diamond lobster	Very strong readiness
4/6	B	🐚 platinum hermit	Good normal PR; ordinary maintainer review
3/6	C	🦐 gold shrimp	Useful, but confidence is limited
2/6	D	🦪 silver shellfish	Proof or implementation needs work
1/6	F	🧂 unranked krab	Not merge-ready
N/A	NA	🌊 off-meta tidepool	Rating does not apply

Overall follows the weaker of proof and patch quality.
Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics.

Workflow

• ClawSweeper keeps one durable marker-backed review comment per issue or PR.
• Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
• A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
• PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
• Maintainers can also comment @clawsweeper review to request a fresh review only.
• Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
• Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
• Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.

History

Review history (1 earlier review cycle)

• reviewed 2026-08-10T07:38:10.406Z sha 782d71c :: found issues before merge. :: [P1] Preserve confirmation before tiered spending actions