agent-environment-retr…
Use when a completed session needs an agent-environment retrospective. Not for an engineering retrospective from telemetry: use engineering-retrospective.
Use when a human asks to compare one closed or stale tracker issue with current reality and mark it done or update it. Don''t use for bulk issue reconciliation or multi-issue sweeps.
$ npx -y skills add OutlineDriven/odin-claude-plugin --skill issue-now --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/issue-nowContext preview
The summary Claude sees to decide when to auto-load this skill.
Use when a human asks to compare one closed or stale tracker issue with current reality and mark it done or update it. Don''t use for bulk issue reconciliation or multi-issue sweeps.
name: issue-now description: 'Use when a human asks to compare one closed or stale tracker issue with current reality and mark it done or update it. Don''t use for bulk issue reconciliation or multi-issue sweeps.' disable-model-invocation: true
| Field | Bound contract | |---|---| | Trigger | A human explicitly asks to compare one closed or stale tracker issue with current reality and mark it done or update it. | | Authority | Remote: closes, reopens, or updates one tracker issue; requires explicit human invocation. Inspect and propose freely; the invoking human's explicit approval of the exact target and consequence precedes any remote mutation. | | Side effect | Close, reopen, or update only the single identified tracker issue; do not mutate other issues, pull requests, project fields, milestones, or repository content. | | Done | After approval and mutation, a fresh tracker read confirms that the issue state and approved content match current reality. |
1. Resolve the supplied reference to exactly one issue. Stop if it is missing, ambiguous, inaccessible, or resolves to a pull request or another tracker object. Done when: one issue is resolved and confirmed not a pull request or other object. 2. Read the issue's current title, body, state, labels, recent discussion, and completion criteria that are available without changing remote state. Done when: the issue's current state and content are captured. 3. Inspect current, directly relevant evidence available from the tracker and referenced project material. Separate observed facts from human claims, and stop rather than infer completion or invent evidence when the comparison cannot be supported. Done when: observed facts and human claims are separated and evidence is gathered or the gap is reported. 4. Compare each material issue claim or completion criterion with that evidence. Classify the issue as complete, still valid as written, stale but actionable, or indeterminate. Done when: the issue has one classification. 5. Derive the semantic-minimum reconciliation for that one issue: close it when the recorded work is demonstrably complete; reopen it when a closed issue remains actionable; otherwise update only the stale fields or text needed to reflect current reality. Preserve accurate history and content. Done when: one reconciliation action is derived that changes the minimum needed. 6. Present a pre-mutation preview naming the issue, its current state, the evidence used, the exact proposed changes, and the consequence. Ask for explicit approval of that preview. Treat silence, general encouragement, or approval of a different proposal as no approval. Done when: the user explicitly approves the exact preview, or the skill stops for lack of approval. 7. If the human changes the proposal, repeat the comparison and preview. Do not widen the operation beyond the identified issue. Done when: the revised preview is approved or the skill stops. 8. After explicit approval, perform only the approved remote mutation on that issue. Do not add unapproved comments, labels, assignments, milestones, or related-item changes. Done when: the approved mutation is performed and the tracker acknowledges it. 9. Read the issue again from the tracker and compare its resulting state and content with the approved preview. Report success only when this fresh read confirms the match. Done when: the fresh read confirms the issue state and content match the approved preview.
Return the issue reference, evidence-based classification, approved change, mutation result, and freshly observed final state. The terminal classification is exactly one of `confirmed`, `awaiting-approval`, `indeterminate`, `blocked`, `failed`, or `unconfirmed`; only `confirmed` means the issue now matches current reality.
Formerly the ODIN Claude Plugin. The repository URL is unchanged. Outline-Driven Development, nicknamed ODIN, is a highly opinionated code-agent skill library: principles-first engineering, surgical editing, and workflow automation, published as installable
Repo: OutlineDriven/odin-claude-plugin
Use when a completed session needs an agent-environment retrospective. Not for an engineering retrospective from telemetry: use engineering-retrospective.
Use when a repo needs agent setup, AGENTS.md added or made lean, CLAUDE.md audited, or agent instructions scored or pruned. Not for remote, credential,…
Use when a human explicitly asks for a full repository agent-compatibility pass returning a scored report with prioritized fixes. Not for tasks that require…
Use when setting up a project, auditing agent command permissions, or asking which read-only bash commands and domains to allow. Not for remote, credential,…
Use when asked to build or review a CLI intended for coding agents and return flag-driven, pipeline-safe, idempotent design advice. Not for running or…
Use when the user asks to make the skills framework work in a new harness, IDE, or CLI. Not for remote, credential, publish, deploy, or irreversible changes.