Plan Amendment Audit

    1

    When a running plan has to change, record the mutation, the reason, and the evidence. Afterward, scrub the log and see how the proposed map became what actually ran.

    $9

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Plan Amendment Audit

    Example session with this skill installed

    The sweep just finished. 45% of the 50 results are in Japanese, but we don't have a Japanese analysis branch. Amend the plan to add one and show the audit log.

    • Read your context and instructions
    • Compiled the plan amendment audit

    I have processed the sweep findings. Because the Japanese language threshold (30%) was exceeded, I am amending the plan to include a specialized discovery branch.

    Plan Amendment: Japanese Market Expansion

    FieldValue
    ChangeNew branch: Japanese social discovery
    ReasonCurrent graph lacks language-specific processing for high-volume Japanese signals identified during initial sweep.
    Evidence45% of 50 verified events (22 items) originated from Japanese-language sources.
    MutationaddNodes: [node_jp_discovery, node_jp_filter], addEdges: [sweep_to_jp_discovery, jp_discovery_to_filter]
    {"seq": 12, "type": "plan_amendment", "change": "New branch: Japanese social discovery", "reason": "Current graph lacks language-specific processing.", "evidence": "45% of 50 verified events from Japanese sources.", "mutation": {"addNodes": [{"id": "node_jp_discovery", "agent": "researcher"}], "addEdges": [{"from": "node_sweep", "to": "node_jp_discovery"}]}}
    

    Next steps

    • Monitor node_jp_discovery for initial fetch success.
    • Run python3 scripts/replay_audit.py events.jsonl --check-amendments to verify log integrity.
    • Verify if the Japanese filter requires additional API keys for localized sentiment analysis.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Agents adapt in their heads and the picture you were shown goes stale. This skill forbids that. If execution needs a node the graph doesn't contain, the agent records a plan amendment with a title, a reason, evidence from real worker output, and the exact change. No evidence, no amendment. The checker re-validates the graph before the change is accepted. When the run ends, the same log is the audit trail: export it, fold it to any point, and draw the map at that moment. Human interventions and agent amendments share the log, so nobody can hide a kill or a pivot. Where it stops: it never rewrites old events. Append only.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Legal & Compliance.

    Structural changes only; no instruction edits. Requires valid DAG mutations. Amendments must be backed by observed evidence, not hunches.

    How to use

    1. 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. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 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 19 days ago

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Listed19 days ago

    What's inside

    Frequently Asked Questions

    This skill prevents 'plan drift' by forcing agents to justify any deviation from their original workflow with measurable evidence, creating a verifiable audit trail instead of letting the agent silently change its behavior.

    The skill includes the JSONL-based event logging schema, the Python replay utility (scripts/replay_audit.py), and the structural mutation protocol for DAG-based workflows.

    It is designed for any agentic framework where plans are represented as Directed Acyclic Graphs (DAGs) and can be integrated into custom Python-based agent loops.

    Yes, the 'timeline scrubbing' feature allows you to use the replay script to visualize the exact state of the plan at any specific sequence number or execution step.

    The skill requires that all mutations result in a valid DAG with unique node IDs and does not currently support modifying existing node instructions—only structural graph changes.

    New in Legal & Compliance