AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Content Review

skill-archiesa-skills-content-review · by ArchieSA

Final read-through of any prose before it ships — a sound pass for AI tells and machine rhythm, an argument pass for whether the claims actually hold, and a register/cohesion pass. Returns specific notes — what's not working and how you'd fix it — not vibes. Use when you say "review this", "read it before I publish", "check for AI tells", or hand any draft for a final pass.

No reviews yet
0 installs
34 views
0.0% view→install

Install

$ agentstack add skill-archiesa-skills-content-review

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No 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.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-archiesa-skills-content-review)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
3mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

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 →
Are you the author of Content Review? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Content review

Three passes over any draft, plus a format check. The output is a list of concrete notes — what specifically fails (the hook, the structure, a claim, filler, the ending) and "here's how I'd do it instead" — never bare "I don't like it."

If a SPEAKER-PROFILE.md or a house-style file is present in the working directory, read against it too: the register, voice, and stop-list it sets are part of the bar.

Pass 1 — how it sounds (AI tells)

The strongest tell isn't a word, it's rhythm and symmetry. Look there first; lexicon second. Full checklist: [AI-TELLS.md](./AI-TELLS.md).

  • Rhythm and symmetry first. Balanced antitheses ("not X, but Y") — one or two is a device, six in a row is a machine. Anaphora chains. Even grids of identical sections. Triads on autopilot.
  • Then lexicon. Filler connectives ("it's worth noting that"), empty intensifiers ("truly", "the very"), nominalized bureaucratese, stock metaphors, summary endings that just restate.
  • Unsupported sweep. "Everyone / most / often / it's accepted that" with no basis — cut it or mark it as opinion.
  • Don't sterilize. Author hedges ("in my experience"), one-off qualifiers, apt jargon, and live "was X → became Y" images carry the voice — touch them last. See AI-TELLS.md, "Don't touch."
  • Final filter: would a person say this out loud?

Pass 2 — what it stands on (the argument)

Run the load-bearing claims — the strongest first — through the argument check. Full checklist: [ARGUMENT-CHECK.md](./ARGUMENT-CHECK.md).

  1. Confidence level marked: fact / experience / hypothesis.
  2. Basis visible — and relevant, sufficient, reliable.
  3. No logical fallacy (ad populum, false dilemma, hasty generalization, correlation-as-cause, straw man, …).
  4. No author bias dressed as fact (confirmation, novelty, survivorship, anchoring).
  5. Steelman done if something is being criticized.
  6. "Where it breaks" names real limits, not a token disclaimer.
  7. No apparent self-contradiction left unexplained (a claim that flips "cheap → not free" must show the scale changed — a unit vs a sum — not the fact).

A found hole gets closed with a basis, softened to a hypothesis, or cut. A disputed technical claim goes to the fact-check skill — not a confident tone.

Pass 3 — register and cohesion

  • One register throughout — forms of address not mixed (check against the profile if there is one).
  • Every block answers "what does this give next"; name the "so what?" blocks that just hang.
  • The title passes the honest-hook test: it promises exactly what the material delivers, no air.

Format check

If the piece has a format with its own checklist (a post, a talk, a lesson, a video), run it against that. A talk, for example: block timings sum to the booked length; the frame and terms don't contradict the CONTEXT file; nothing repeats a neighboring talk in the series. Keep format rules in the material or its skill, not here — this skill is format-agnostic.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.