How do I install the Report Smoke skill?

Run npx skillmds@latest add zapgun-ai/report-smoke in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Report Smoke safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Report Smoke?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Report Smoke free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Report Smoke?

zapgun-ai (@zapgun-ai) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Report Smoke by zapgun-ai — Agent Skill

Name: Report Smoke
Author: zapgun-ai

Report Smoke

End-to-end smoke test for the /<admin-path>/report saved-run viewer. Boots the real clawback binary against the repo's runs/ dir and curls every report route, asserting status + content, prefix mutability, public-read exemption, and that sensitive run-dir siblings stay unreachable. Use after touching src/report.js, src/report_ui/*, or the admin/report wiring.

zapgun-ai Updated Sep 17, 2026

File contents

clawback report viewer smoke test

Run .skills/report_smoke/scripts/report_smoke.sh from the project root:

.skills/report_smoke/scripts/report_smoke.sh

Environment variables (optional):

CLAWBACK_REPORT_DIR — runs directory to serve (default ./runs)

It needs at least one completed run on disk (a subdir with a parseable summary.json). The repo ships runs/smoke and runs/L0-tier1; the L0-tier1 run also contains the sensitive proxy.*.log + turns.*.ndjson siblings, which is exactly what the non-exposure assertions check against.

No upstream and no API key are required — report routes never proxy to Anthropic. The proxy boots over plain HTTP on a free high port and is torn down on exit. Exit code is the number of failed assertions (0 = all green).

What the script verifies:

Static + base injection — /_proxy/report/ serves index.html with __BASE__ replaced by the real /_proxy/report/ base href (no literal __BASE__ leaks); report.js/report.css serve with correct MIME types.
Dynamic routes against real analyzer output — /runs lists the on-disk runs newest-first; /data?run=L0-tier1 returns summary (carrying the tokens reclaim block) + charts + csvBytes; /chart/<id>/tokens_saved.svg serves that SVG by name (the same route also serves the bare tokens_saved.bg.svg share-card background); /csv/<id> serves a CSV download.
Security (allowlist is the control; traversal guards are depth) — the proxy.*.log / turns.*.ndjson siblings are unreachable, the data payload does not leak their bytes, and ?run=.. / encoded traversal are rejected 400.
Publicly readable — a GET with a bogus Host 421s on a guarded endpoint (/_proxy/metrics) but the report viewer is exempt and still returns 200.
Dashboard cross-link — /_proxy/ui/ exposes the reportLink.
Prefix mutability — a re-boot with --admin-path ctrl serves /ctrl/report/... (base href injected as /ctrl/report/) while the old /_proxy/report/ path 404s, proving the admin prefix is not hardcoded.

Complements test/report.test.js (unit-level, synthetic run dirs): this exercises the shipped bin/clawback.js boot path against real runs.

---name: report-smokedescription: End-to-end smoke test for the /<admin-path>/report saved-run viewer. Boots the real clawback binary against the repo's runs/ dir and curls every report route, asserting status + content, prefix mutability, public-read exemption, and that sensitive run-dir siblings stay unreachable. Use after touching src/report.js, src/report_ui/*, or the admin/report wiring.---# clawback report viewer smoke testRun `.skills/report_smoke/scripts/report_smoke.sh` from the project root:```bash.skills/report_smoke/scripts/report_smoke.sh```Environment variables (optional):- `CLAWBACK_REPORT_DIR` — runs directory to serve (default `./runs`)It needs at least one completed run on disk (a subdir with a parseable`summary.json`). The repo ships `runs/smoke` and `runs/L0-tier1`; the L0-tier1run also contains the sensitive `proxy.*.log` + `turns.*.ndjson` siblings, whichis exactly what the non-exposure assertions check against.No upstream and no API key are required — report routes never proxy toAnthropic. The proxy boots over plain HTTP on a free high port and is torn downon exit. Exit code is the number of failed assertions (0 = all green).What the script verifies:1. **Static + base injection** — `/_proxy/report/` serves index.html with `__BASE__` replaced by the real `/_proxy/report/` base href (no literal `__BASE__` leaks); `report.js`/`report.css` serve with correct MIME types.2. **Dynamic routes against real analyzer output** — `/runs` lists the on-disk runs newest-first; `/data?run=L0-tier1` returns summary (carrying the `tokens` reclaim block) + charts + csvBytes; `/chart/<id>/tokens_saved.svg` serves that SVG by name (the same route also serves the bare `tokens_saved.bg.svg` share-card background); `/csv/<id>` serves a CSV download.3. **Security (allowlist is the control; traversal guards are depth)** — the `proxy.*.log` / `turns.*.ndjson` siblings are unreachable, the data payload does not leak their bytes, and `?run=..` / encoded traversal are rejected 400.4. **Publicly readable** — a GET with a bogus `Host` 421s on a guarded endpoint (`/_proxy/metrics`) but the report viewer is exempt and still returns 200.5. **Dashboard cross-link** — `/_proxy/ui/` exposes the `reportLink`.6. **Prefix mutability** — a re-boot with `--admin-path ctrl` serves `/ctrl/report/...` (base href injected as `/ctrl/report/`) while the old `/_proxy/report/` path 404s, proving the admin prefix is not hardcoded.Complements `test/report.test.js` (unit-level, synthetic run dirs): thisexercises the shipped `bin/clawback.js` boot path against real runs.

Frequently asked questions