Write Goal
Phase 2 of pair-goal. Turns .pairgoal/UNDERSTANDING.md into a /goal-ready completion condition — one measurable end state, a stated transcript-demonstrable check, the constraints that must hold, and a turn cap — written to .pairgoal/GOAL.md plus the literal /goal line for the user to run. Use after train-orchestrator, or standalone to craft a good /goal condition from a clear improvement target.
Install
$ agentstack add skill-ccomkhj-skills-write-goal ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Links to your public security report.[](https://agentstack.voostack.com/security/report/skill-ccomkhj-skills-write-goal) Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
write-goal
Phase 2 of pair-goal. Converts .pairgoal/UNDERSTANDING.md into a completion condition you can hand to Claude Code's /goal command, written to .pairgoal/GOAL.md.
What makes a good /goal condition
/goal's evaluator is a fast model that, after every turn, reads only the conversation (it runs no tools) and decides yes/no. So a strong condition has all four parts:
- One measurable end state — a test result, a benchmark threshold, a file count, an empty queue, an exit code. Not "better" or "cleaner".
- A stated check — how it's proven, phrased so the agent's own output demonstrates it:
pytest tests/x -q exits 0, `bench/run.py prints p95 Goal condition ready. Review it, then run this yourself to start the loop:
> > /goal > > Once it's active, each turn I'll run one pair-iterate race round until the > condition holds, then summarize.
Set STATUS: WAITING-USER: approve-goal. Wait for the user to approve (and possibly edit) and to actually run /goal. Only then does pair-iterate begin.
Don't
- Don't make the condition depend on something the agent won't print. If proving it requires reading a file or running a tool, make the agent run that and surface the output part of the loop's job — and say so in GOAL.md.
- Don't omit the turn clause. A goal with no cap can run away.
- Don't widen scope beyond UNDERSTANDING.md. The goal is the contract; keep it tight.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: ccomkhj
- Source: ccomkhj/skills
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.