Agent Reliability Kit
Seven guards for AI coding agents, each built after a real failure in our own production system. They stop the agent saying "not found" from a partial read, "done" when it only pushed, printing secrets, trusting stale memory notes, losing the last session's work, breaking the phone layout, and citing sources that do not say what it claims.
Works with any agent that signs in to Agensi, such as Claude, ChatGPT, Claude Code, Cursor and Codex CLI.
One payment, lifetime access. 7 skills unlock instantly in your library.
What's included
7 skillsThree scripts that refuse a clean verdict from a partial read: complete sweeps with N of N, PDF page coverage, and a lint for unbacked "there is no X" claims.
Checks a summary or deal brief claim by claim against the source document and lists what is supported, overstated, missing or changed.
A PreToolUse hook that refuses, before it runs, any command that would print a secret into the transcript (shell tracing, env dumps, reading .env files, docker inspect, echoing key variables), while the safe patterns still pass. 94-case selftest; 6 of 7,454 real commands refused.
Reads your Claude Code transcripts and shows, newest session first, what you typed and what the agent actually ran (commits, pushes, installs, deploys, sends), then blocks "ready to close" while repos are dirty or unpushed or the NEXT SESSION note is missing.
Checks the [verify | checked: date] tags in CLAUDE.md, AGENTS.md, notes and skills (too old, malformed, future-dated), lists every file read N of N and exits 2, so stale memory cannot reach your agent's answers.
Reads your change back from the live URL (text present, old text gone, or byte-for-byte hash) next to a nonsense-path control, so a catch-all 200 never passes as proof. LIVE / NOT LIVE / COULD NOT VERIFY with exit codes; 18-case selftest on local servers.
Renders every page at phone width and measures each element against the screen edge, so overflow hidden by overflow-x:hidden is still caught. Clean / overflow / could not check, with exit codes.
See it in action
False Negative Guard: Prove 'Not Found' Before You Say It
Example session with this skill installed
Check whether any of our repositories has a pending collaborator invitation, and tell me if the answer is "none".
- Read your context and instructions
- Compiled the false negative guard
Summary
The guard swept every repository instead of the first page of results, and found one pending invitation. So the honest answer is not "none".
! delta -> invitee: client-x, expires 2026-10-19
pending invitations: examined 5 of 5 item(s): 1 hit(s), 4 clean, 0 UNKNOWN
FOUND. The items above are the evidence.
Answer: one invitation is pending on delta, to client-x, expiring 2026-10-19.
The same guard, run on a draft reply before sending it
negative-claim lint: 2 negative claim(s) carry no evidence in their paragraph:
- There is no pending invitation on any repository, so nothing blocks the handover.
- The tender documentation is incomplete: the technical specification is not in the document.
Each needs one of: 'examined N of N', 'COMPLETE SWEEP', 'READ COMPLETE', or '[verified: <surface>, N of N]' in the same paragraph, or a stated scope ('in the 12 files I read').
On a real 29-page procurement PDF whose first 12 pages were scans
An agent once told a public authority its specification was missing. It was on the scanned pages.
page 1: 0 chars <- IMAGE-ONLY (needs OCR)
page 2: 0 chars <- IMAGE-ONLY (needs OCR)
page 12: 0 chars <- IMAGE-ONLY (needs OCR)
...
page 13: 456 chars
page 14: 2969 chars
page 29: 882 chars
coverage: 17 of 29 page(s) have a text layer (58%)
NOT READ: 12 page(s) are image-only: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12. OCR them before saying anything is absent from this document.
--grep 'specifica': 2 match(es) over 17 readable page(s) of 29.
Next steps
- Run each script's --selftest once (21 cases) to see the exit codes on your machine.
- Put sweep.py in front of any "none / nothing / all clear" answer your agent gives.
- Run negative_claim_lint.py on the agent's final reply before you trust it.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads these skills and uses them on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every skill in this bundle goes through our trust checks.
Security scanned
Checked by our 8-point scan before listing
10 uses
Times these skills were used to date
Money-back guarantee
Doesn't do what its page promises? You get your money back.