feat(skills): add Astra to Sol delegation workflow

[notorious-d-e-v]

Add an Astra-to-Sol counterpart to codex-first: in Codex sessions using GPT-6 Astra, delegate bounded implementation or exploration to GPT-5.6 Sol at the reasoning effort Astra chooses, then have Astra review and verify the result. The existing Claude-to-Codex workflow is unchanged.

The skill prefers native subagents with an explicit Sol model and fresh context, includes a CLI fallback with standard service and scoped execution, and handles unavailable models without silent substitution. It follows skills/<name>/SKILL.md, with optional Codex UI metadata and a bundled MIT notice so it can be installed independently.

Also distributed in notorious-d-e-v/agent-skills, a collection using the same folder layout.

Validation

• scripts/validate-skills: validated 55 skills in this checkout.
• Codex skill-creator validator: valid skill.
• git diff --check: passed.
• Real download/install from the public collection using Codex's bundled skill-installer: passed; installed files match the source.
• Isolated Sol autoreview at high reasoning, P0–P2 scope: no actionable findings.

Validation covers packaging and the documented launch contract. Live delegation across other Codex versions/accounts is not claimed; model access and tool support are explicit prerequisites.

[clawsweeper]

🦞👀
ClawSweeper picked this up.

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

ClawSweeper review complete

ClawSweeper finished reviewing this revision. The review result is being finalized.

View the workflow run.

[clawsweeper]

Codex review: needs real behavior proof before merge. Reviewed September 29, 2026, 11:54 AM ET / 15:54 UTC (Revision 2).

ClawSweeper review

What this changes

Adds an installable Codex skill that guides Astra to assign bounded work to Sol, choose reasoning effort, and review the worker's result.

Merge readiness

⛔ Blocked before merge - 5 items remain

This is a distinct skill that current main does not provide. It remains useful, but implicit Sol routing conflicts with the repository's pinned worker recipe, and the submitted validation does not show a completed Astra-to-Sol delegation.

Priority: P3
Reviewed head: a1107e3af880468a9a34479427ea522df25dd72f
Owner decision: Required. See Decision needed.

Review scores

Measure	Result	What it means
Overall readiness	🦪 silver shellfish (2/6)	The skill is focused and package validation is reported, but the implicit-routing conflict and missing delegation trace prevent merge confidence.
Proof confidence	🦪 silver shellfish (2/6)	Needs stronger real behavior proof before merge: The PR reports real installation of the skill, but supplies no after-install Codex run showing a scoped Sol worker, its chosen effort, completed result, and Astra verification. A redacted terminal transcript or recording of that path would address the previous review request; redact private endpoints, identifiers, and credentials before posting. Updating the PR body should trigger re-review, or a maintainer can request @clawsweeper re-review. No stored-data contract changes. After adding proof, update the PR body; ClawSweeper should re-review automatically. If it does not, the PR author or someone with repository write access can comment @clawsweeper re-review.
Patch quality	🦐 gold shrimp (3/6)	1 actionable review finding remain.

Verification

Check	Result	Evidence
Real behavior	Needs proof	Needs stronger real behavior proof before merge: The PR reports real installation of the skill, but supplies no after-install Codex run showing a scoped Sol worker, its chosen effort, completed result, and Astra verification. A redacted terminal transcript or recording of that path would address the previous review request; redact private endpoints, identifiers, and credentials before posting. Updating the PR body should trigger re-review, or a maintainer can request @clawsweeper re-review. No stored-data contract changes. After adding proof, update the PR body; ClawSweeper should re-review automatically. If it does not, the PR author or someone with repository write access can comment @clawsweeper re-review.
Evidence reviewed	7 items	Introduced routing: The new skill applies in Astra sessions, selects Sol workers, and supplies a standard-service CLI fallback. Implicit invocation: The added metadata permits the new routing skill to be invoked implicitly. Repository worker policy: The repository's canonical uppercase AGENTS.MD requires Codex worker model, effort, and service tier to follow the codex-first recipe unless the user requests an override.
Findings	1 actionable finding	[P1] Make Sol routing an explicit user choice
Security	None	None.

How this fits together

The repository's skills are discovered from installed skill folders and guide agent behavior during a session. This skill takes an Astra task, routes a bounded assignment to a Sol worker through collaboration tools or the Codex CLI, and returns the result for Astra to verify.

flowchart LR
  A[Astra session] --> B[Installed skill]
  B --> C{Delegate this task?}
  C -->|Yes| D[Sol worker]
  C -->|No| E[Astra works directly]
  D --> F[Worker result]
  F --> G[Astra review and verification]

Loading

Decision needed

Question	Recommendation
Should this repository permit Astra sessions to route work to Sol at standard service implicitly, or require an explicit user override of the pinned worker recipe?	Require explicit opt-in: Keep the pinned recipe for ordinary sessions and invoke the Sol skill only when the user explicitly chooses it.

Why: The proposed implicit behavior conflicts with the repository's current hard routing rule; choosing a new default requires owner intent.

Before merge

• Add real behavior proof - Needs stronger real behavior proof before merge: The PR reports real installation of the skill, but supplies no after-install Codex run showing a scoped Sol worker, its chosen effort, completed result, and Astra verification. A redacted terminal transcript or recording of that path would address the previous review request; redact private endpoints, identifiers, and credentials before posting. Updating the PR body should trigger re-review, or a maintainer can request @clawsweeper re-review. No stored-data contract changes. After adding proof, update the PR body; ClawSweeper should re-review automatically. If it does not, the PR author or someone with repository write access can comment @clawsweeper re-review.
• Make Sol routing an explicit user choice (P1) - The new metadata permits implicit invocation, which can route an ordinary Astra session to Sol at standard service. AGENTS.MD requires Codex workers to follow the existing Astra/high/Fast recipe unless the user requests an override. Disable implicit invocation or obtain and record a policy change before this skill can alter that default. This concern was missed in the earlier review of the unchanged head.
• Resolve merge risk (P1) - With implicit invocation enabled, an existing Codex session may select Sol and standard service without the explicit user override required by the repository's pinned worker policy.
• Complete next step (P2) - Resolve whether Sol routing must be explicit, update the skill or shared policy to match that decision, and add a redacted real Codex delegation trace before merge.
• Resolve maintainer decision - Resolve the maintainer decision shown above before merge.

Findings

• [P1] Make Sol routing an explicit user choice — skills/astra-sol-delegation/agents/openai.yaml:6

Agent review details

Security

None.

Review metrics

None.

Merge-risk options

Maintainer options:

1. Make routing opt-in (recommended)
   Disable implicit invocation and state that selecting Sol overrides the repository's normal worker recipe.
2. Adopt the new default
   Approve and update the shared routing rule so installed sessions have one coherent model and service policy.

Technical review

Best possible solution:

Keep the existing worker recipe as the default, make Sol routing an explicit opt-in if approved, and verify one installed skill run from assignment through Astra's final review.

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

Not applicable: this PR proposes a new optional skill, and its runtime behavior has not been demonstrated in the submitted evidence.

Is this the best way to solve the issue?

No: the skill fills a distinct Codex workflow, but implicit invocation is not the safest fit with the repository's current worker policy. Explicit opt-in is the narrower path.

Full review comments:

• [P1] Make Sol routing an explicit user choice — skills/astra-sol-delegation/agents/openai.yaml:6
  The new metadata permits implicit invocation, which can route an ordinary Astra session to Sol at standard service. AGENTS.MD requires Codex workers to follow the existing Astra/high/Fast recipe unless the user requests an override. Disable implicit invocation or obtain and record a policy change before this skill can alter that default. This concern was missed in the earlier review of the unchanged head.
  Confidence: 0.91
  Late finding: first raised on code an earlier review cycle already covered.

Overall correctness: patch is incorrect
Overall confidence: 0.89

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning medium; reviewed against 3f8c6a33f911.

Labels

Label changes:

No label changes.

Label justifications:

• P3: This optional workflow skill is a low-urgency improvement rather than a current user-facing outage.
• merge-risk: 🚨 compatibility: Implicit invocation can replace the repository's established worker model, effort, and service tier without a user override.
• rating: 🦪 silver shellfish: Overall readiness is 🦪 silver shellfish; proof is 🦪 silver shellfish and patch quality is 🦐 gold shrimp.
• status: 📣 needs proof: The PR needs real behavior proof before ClawSweeper can clear the contributor ask. Needs stronger real behavior proof before merge: The PR reports real installation of the skill, but supplies no after-install Codex run showing a scoped Sol worker, its chosen effort, completed result, and Astra verification. A redacted terminal transcript or recording of that path would address the previous review request; redact private endpoints, identifiers, and credentials before posting. Updating the PR body should trigger re-review, or a maintainer can request @clawsweeper re-review. No stored-data contract changes. After adding proof, update the PR body; ClawSweeper should re-review automatically. If it does not, the PR author or someone with repository write access can comment @clawsweeper re-review.

Evidence

What I checked:

• Introduced routing: The new skill applies in Astra sessions, selects Sol workers, and supplies a standard-service CLI fallback. (skills/astra-sol-delegation/SKILL.md:11, a1107e3af880)
• Implicit invocation: The added metadata permits the new routing skill to be invoked implicitly. (skills/astra-sol-delegation/agents/openai.yaml:6, a1107e3af880)
• Repository worker policy: The repository's canonical uppercase AGENTS.MD requires Codex worker model, effort, and service tier to follow the codex-first recipe unless the user requests an override. (AGENTS.MD:41, a1107e3af880)
• Current recipe: Current main's codex-first launch recipe pins Astra, high reasoning, and Fast service for workers. (skills/codex-first/SKILL.md:5, 3f8c6a33f911)
• Still distinct on main: The fetched current main tree has no astra-sol-delegation skill directory; codex-first explicitly gates its hands-on routing to Claude Code sessions. (skills/codex-first/SKILL.md:18, 3f8c6a33f911)
• Prior proof request remains open: The PR body reports package validation and a real installation, but no observed Sol assignment and Astra verification. The previous ClawSweeper review requested a redacted runtime trace, and the reviewed head remains unchanged. (a1107e3af880)

Likely related people:

• Peter Steinberger: Suggested for follow-up; no historical authorship or introduction is verified. (role: unverified routing candidate; confidence: low)
• notorious-d-e-v: Suggested for follow-up; no historical authorship or introduction is verified. (role: unverified routing candidate; confidence: low)

Rank-up moves

Optional improvements that raise the rating; they are not merge blockers.

• Resolve the implicit-routing policy conflict through explicit opt-in or an approved policy update.
• Add redacted runtime output showing the installed skill's worker model and effort, completed assignment, and parent verification in an identified Codex version.

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 2026-09-15T03:39:33.134Z sha a1107e3 :: needs real behavior proof before merge. :: none

[steipete]

Thanks for the thoughtful packaging and validation. I’m closing this because model routing here is owned by codex-first and the active session’s delegation policy; a second implicitly invoked, model-specific routing skill would create a competing policy and service-tier default. Keeping this as an independently installable skill is a better fit than adding it to this repository’s default skill set.