issue-verify
Re-reads open issues against the current tree and posts a verification comment on each, flagging anything already fixed, partly stale, or wrong to do.
Skill by philoserf
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/philoserf-issue-verify
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md and code. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
- Runs as
- Shell scripts
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/issue-verify
What's inside
- Skill name
issue-verify- Description
- Re-reads open issues against the current tree and posts a verification comment on each, flagging anything already fixed, partly stale, or wrong to do. Use before acting on a backlog, triaging a milestone, or when asked whether filed issues are still real.
- License
- MIT
- Instructions
- 94 lines in SKILL.md · view on GitHub
Before installing
- Ships executable scripts
Agent Skills spec issues
- name: missing
- allowed-tools: must be a space-separated string
Files (2)
- SKILL.md
- issue-facts.sh · scripts/
SKILL.md (first lines)
# Issue verification An issue is a claim about a tree that has since moved. This skill re-reads each one against current `HEAD`, decides whether the claim still holds, and writes that decision down where the next reader will find it — on the issue. Verification is not agreement. The output is a disposition for every issue looked at, and "this was fixed three commits ago" is as useful a result as "still stands". ## Gather the facts first ```bash ~/.claude/skills/issue-verify/scripts/issue-facts.sh <N> [N...] ``` Per issue it prints the metadata, then rows you can grep: ```text FILE present|MISSING <path> every source path the body names SYMBOL present|MISSING <name> every backticked identifier SINCE filed at <sha>; <n> commit(s) touched its files since DEP blocked_by #N [OPEN|CLOSED] ``` **`SINCE` is the row that earns this script.** It lists the commits that have touched the issue's own files since it was filed, which is the difference between an issue nobody has been near and one whose ground has shifted under it. **A `MISSING` row is a prompt, not a verdict.** An issue may correctly name a file it is asking you to create, or a symbol it is asking you to delete. Read before concluding. ## Then read the issue The script cannot tell you whether a finding is _right_, only whether its references still resolve. Open the body and check its claims against the code — the counts, the greps, the line it quotes. Findings written by the `code-*` skills state how they were verified; redo that, do not take it on trust. Four dispositions, and every issue gets exactly one: - **Stands.** The claim is still true. Say what you re-checked. - **Already fixed.** Name the commit that did it and close. - **Partly stale.** The most common and the easiest to miss: half the finding has been overtaken and the rest is live. Say which half is gone, narrow the scope, keep the issue. - **Wrong to do.** The suggested change would make things worse, or the premise has been invalidated. Argue it on the issue rather than closing quietly. ## Verify the claims the issue could not The highest-value finding in a verification pass is usually a claim the author marked unverifiable. `code-*` findings are written offline and say so — _"I cannot verify what USNO publishes for that date"_, _"I could not check this against the upstream"_. **Try.** Fetch the reference, run the command, query the API. The answer can change the correct fix rather than confirm it — published data may show the value right and the _comment_ wrong, which calls for a tighter test rather than a weaker label. If it stays unverifiable, say that in the comment, with what you tried. ## Watch for - **Issues overtaken by a deletion.** A correctness finding inside code another issue proposes to delete does not need fixing; it needs closing with the deletion. Check whether a sibling issue removes the ground this one stands on. - **`duplicate` labels.** GitHub records the relation as `blocked_by`; the script prints it. A duplicate closes on the other issue's PR, not by itself. Where the relation is described in prose but was never recorded, `issue-triage` is what records it. - **Half-stale infrastructure findings.** An issue naming two files where one has since been deleted is still a real finding about the survivor. Narrow it; do not close it. - **A body that references a file the repo hides.** `.issues/` is excluded by a global ignore in some repos, so its paths report MISSING by design. The script already drops the `Filed from` footer's own path. ## Post one comment per issue Say what you re-checked, what you found, and the disposition. Quote the evidence — a grep count, a commit sha, a line that no longer exists. A verification comment that says only "still valid" has not recorded anything the next reader could not have guessed. Where the disposition changes the scope, say so explicitly, because the issue title will keep claiming the original scope forever. ## Do not use when - You are about to fix the issue in the same pass — use `issue-pr`, which verifies as its first step and does not need a separate comment. - Nothing has landed since the issues were filed. `SINCE` will say so, and a pass that re-confirms untouched findings is noise on every issue. - The issues are fine but unsorted — no milestone, no relations. That is placement rather than truth: use `issue-triage`.
From philoserf/claude-code-config, MIT.
More from philoserf/claude-code-config
- 0.641score
gh-stack
github
Manages stacked PRs and splits multi-part work into reviewable branches with gh-stack.
SkillAFreshness: A★ 13Claude Code +9 - 0.637score
editor
philoserf
Copy edits prose while preserving voice and register.
SkillAFreshness: A★ 13Claude Code +9 - 0.633score
cc-release-review
philoserf
Reviews Claude Code release notes and recommends config updates.
SkillAFreshness: A★ 13Claude Code +9 - 0.633score
issue-pr
philoserf
Fixes one issue and lands it as a merged pull request — branch, implement, run the repo's own gate, open a PR that closes the issue, merge.
SkillAFreshness: A★ 13Claude Code +9 - 0.633score
issue-triage
philoserf
Organizes filed issues into milestones and repairs their metadata — severity labels and blocked-by relations.
SkillAFreshness: A★ 13Claude Code +9 - 0.633score
release-gate
philoserf
Validates a repository is ready to tag and ship.
SkillAFreshness: A★ 13Claude Code +9 - 0.633score
release-ship
philoserf
Executes the release workflow for a repository after release-gate passes.
SkillAFreshness: A★ 13Claude Code +9 - 0.633score
code-audit
philoserf
Reviews a codebase for bugs, design issues, and code cleanliness problems with specific file paths and line numbers.
SkillAFreshness: A★ 13Claude Code +9 - 0.631score
grill
philoserf
Interviews the user relentlessly about a plan or design until every decision in it is resolved.
SkillAFreshness: A★ 13Claude Code +9 - 0.630score
release-captain
philoserf
Works an entire milestone unattended: verifies every open issue, sequences them by dependency, ships each as a merged PR, then stops when the milestone is…
SkillAFreshness: A★ 13Claude Code +9 - 0.628score
code-theory
philoserf
Builds a Naur-style theory of a codebase — the understanding needed to change it without damaging its conceptual integrity.
SkillAFreshness: A★ 13Claude Code +9 - 0.628score
code-walkthrough
philoserf
Reads source code and produces a linear walkthrough document.
SkillAFreshness: A★ 13Claude Code +9
Signals
- Stars
- 13
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Oct 12, 2026 (16 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/philoserf/claude-code-config
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from philoserf
- 0.637score
editor
philoserf
Copy edits prose while preserving voice and register.
SkillAFreshness: A★ 13Claude Code +9 - 0.633score
release-gate
philoserf
Validates a repository is ready to tag and ship.
SkillAFreshness: A★ 13Claude Code +9 - 0.633score
cc-release-review
philoserf
Reviews Claude Code release notes and recommends config updates.
SkillAFreshness: A★ 13Claude Code +9 - 0.633score
issue-pr
philoserf
Fixes one issue and lands it as a merged pull request — branch, implement, run the repo's own gate, open a PR that closes the issue, merge.
SkillAFreshness: A★ 13Claude Code +9 - 0.633score
issue-triage
philoserf
Organizes filed issues into milestones and repairs their metadata — severity labels and blocked-by relations.
SkillAFreshness: A★ 13Claude Code +9 - 0.633score
release-ship
philoserf
Executes the release workflow for a repository after release-gate passes.
SkillAFreshness: A★ 13Claude Code +9
Cite this page
issue-verify skill. Agent Mart. https://agentmart.dev/skills/philoserf-issue-verify (last updated 2026-10-24)