More screenshots
Works with the AI Agents you already use
AI Tool Call Chain Integrity & Cost Receipt
Verify local AI tool-call and output logs for missing pairs, schema drift, duplicate calls, error outcomes and supplied cost-limit breaches.
$9
AI Tool Call Chain Integrity & Cost Receipt
Example session with this skill installed
Review my authorized local exports tool-calls.jsonl and tool-outputs.jsonl against tool-catalog.json and cost-limits.json. Create a new packet in ./tool_call_review, identify missing pairs, duplicate calls, schema drift, error outcomes and supplied cost-limit breaches.
- Read your context and instructions
- Compiled the ai tool call
- Generated the data export
Tool-call integrity review complete
The two exported logs were joined locally by exact call ID. The source files were not modified, and evidence-receipt.json binds all inputs and generated outputs with SHA-256.
| Check | Sample result |
|---|---|
| Calls inspected | 86 |
| Exact call/output pairs | 82 |
| Missing outputs | 2 |
| Orphan outputs | 1 |
| Argument-schema findings | 3 |
| Duplicate fingerprint groups | 2 |
| Error outcomes | 4 |
| Supplied usage cost | $18.42 of $20.00 limit |
Findings requiring review
- call_041 and call_078 have no matching output record.
- output_legacy_02 refers to a call ID absent from the supplied call log.
- Three calls include undeclared argument fields under the supplied tool catalog.
- Two repeated-call groups share normalized tool name and arguments; they are reported as candidates and were not automatically treated as billing duplicates.
- Four matched outputs declare an error status.
Files created
call-matrix.csv provides one row per joined record, duplicate-call-groups.csv groups repeated fingerprints, cost-summary.csv totals the supplied costs, issues.jsonl stores findings, and TOOL_CALL_INTEGRITY.md provides the human review. result.json and evidence-receipt.json preserve totals and hashes.
Next steps
Investigate the missing pairs and repeated groups using the original run context. Confirm costs against the vendor invoice if billing accuracy matters. The checker did not execute tools, recount tokens, grade output meaning or prove that any logged action occurred.
call-matrix.csv
CSV · data export
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
This downloadable local package turns authorized buyer-supplied AI tool-call and output logs into a fixed, reproducible review packet. It includes an installable Python CLI, a local stdio MCP adapter and a companion Agent Skill. One run creates:
- result.json
- call-matrix.csv
- duplicate-call-groups.csv
- cost-summary.csv
- issues.jsonl
- TOOL_CALL_INTEGRITY.md
- evidence-receipt.json
The deterministic core joins calls to outputs by exact call ID, checks declared tool names and argument fields against a supplied catalog, detects missing and duplicate evidence, groups repeated call fingerprints and totals supplied usage costs under explicit limits. It rejects unsafe paths and malformed inputs, never overwrites an existing packet, escapes formula-active CSV cells and binds every input and output with SHA-256. Processing stays local; no API key, login, upload, telemetry or external execution is included.
Best for AI application teams and integration consultants reviewing exported agent runs. Requires Windows 11, Python 3.11–3.14 and authorized local evidence files. Optional mcp==2.2.0 enables the local adapter.
Limitations: no model or tool execution, trace collection, semantic output grading, token recount, vendor billing verification, retry, rollback or claim that logged actions occurred. This is a one-time local V1.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for LLM Ops.
Requires local JSON logs for calls, outputs, and catalogs. Cannot perform live API calls or remote actions. Analysis is restricted to provided evidence files.
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 this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 2 months ago
- Software
Python
Permissions
Runs the bundled local Python CLI, reads only buyer-selected exported log and policy files, and writes a new review packet locally. No browser, network access or environment secrets required.