chrissaautomates.com

Last change

None yetUnchanged since first read.

Declared read-only

10 of 18

llms.txt

Present

1. add_research_candidateCan change things

   Add a candidate discovered by the agent. Candidates are deduplicated by normalized domain when available, otherwise by normalized name. The agent may add candidates but may not change human-controlled requirements.

   First seen 11 Oct 2026. Unchanged since.

2. approve_firecrawl_escalationCan change things

   Sensitive paid action. Only call after the human explicitly approves paid retrieval. The exact confirmation phrase is exchanged for a short-lived server-signed token scoped to the current contract and URLs before Firecrawl can run.

   First seen 11 Oct 2026. Unchanged since.

3. build_verified_shortlistCan change things

   Use for find, compare, recommend, shortlist, vendor-selection, or product-selection requests where the user has requirements. Discover candidates automatically, verify must-have claims with current web evidence, reject unsupported options, and update the same shortlist visible to the human. URLs are not required from the user.

   First seen 11 Oct 2026. Unchanged since.

4. check_action_clearanceReads only

   Return ACTION_ALLOWED only when every required claim in the current evidence contract is verified; otherwise return ACTION_BLOCKED with the exact unresolved gates.

   First seen 11 Oct 2026. Unchanged since.

5. create_evidence_contractCan change things

   Create the live Agent Evidence Gate state that defines what a vendor recommendation must prove before the agent can proceed. Updates the same state visible to the human.

   First seen 11 Oct 2026. Unchanged since.

6. create_research_boardCan change things

   Create the board visible to the human. At creation time the agent may translate the user's request into must-have criteria. After creation, humans control the rules; agents control candidate discovery and evidence.

   First seen 11 Oct 2026. Unchanged since.

7. get_agent_utility_index_evidenceReads only

   Return the public frozen benchmark evidence that powers evidence-acquisition decisions.

   First seen 11 Oct 2026. Unchanged since.

8. get_evidence_decision_reportReads only

   Return the machine-readable audit report with claim evidence, source URLs, retrieval routes, paid-call counts, action clearance, and frozen benchmark context.

   First seen 11 Oct 2026. Unchanged since.

9. get_missing_checksReads only

   Return only the automatic research queue. MISSING and PARTIAL cells are eligible for automatic recheck. FAILED is terminal and excluded unless the human changes the rule or explicitly asks the agent to revisit it.

   First seen 11 Oct 2026. Unchanged since.

10. get_research_boardReads only

    Read the exact current board state. Pass since_revision when you already saw an earlier revision; the response includes only the recorded changes since then as changes_since, while still returning the current board for safe synchronization.

    First seen 11 Oct 2026. Unchanged since.

11. get_verified_shortlistReads only

    Read the shortlist and requirements currently visible to the human, including qualified, near-match, rejected options and their evidence checks.

    First seen 11 Oct 2026. Unchanged since.

12. inspect_evidence_gapsReads only

    Read the required claims that still block the current agent action.

    First seen 11 Oct 2026. Unchanged since.

13. recommend_agent_retrieval_strategyReads only

    Compute an evidence-guided web retrieval strategy from the frozen Agent Utility Index benchmark without calling a provider.

    First seen 11 Oct 2026. Unchanged since.

14. recommend_evidence_retrievalReads only

    Apply the current contract plus the frozen Agent Utility Index policy to decide whether the agent should FREE_STOP, request Firecrawl approval, or execute an approved escalation.

    First seen 11 Oct 2026. Unchanged since.

15. refine_verified_shortlistCan change things

    Use after the human changes a requirement, such as adding a budget, removing credit-card requirements, changing a minimum allowance, or narrowing a feature. Re-run discovery and verification and update the same visible shortlist.

    First seen 11 Oct 2026. Unchanged since.

16. research_and_verify_web_questionReads only

    Use this when the user needs current web research, vendor discovery, product comparison, pricing/plan verification, feature verification, or an evidence-backed shortlist. The user does not need to provide URLs. Discover candidates, prefer first-party evidence for hard vendor claims, and return the supported answer with sources.

    First seen 11 Oct 2026. Unchanged since.

17. run_direct_evidence_checkCan change things

    Retrieve the contract's first-party URLs with Direct Fetch, test the hard evidence gates, update the visible workspace, and block or clear the action without spending Firecrawl credits.

    First seen 11 Oct 2026. Unchanged since.

18. update_candidate_evidenceCan change things

    Write one researched candidate/requirement cell. Supply the board revision research started from and the exact evidence cell revision expected. Unrelated board edits do not invalidate the write; a changed requirement or newer write to this same cell does. VERIFIED requires a source URL. A rejected VERIFIED-without-source attempt is also written to shared activity so the human can see the guardrail

    First seen 11 Oct 2026. Unchanged since.

Run this site? This page shows what your site told our crawler, ToolChronicleBot. To be left out, disallow it in robots.txt or write to hello@toolchronicle.com.