Tools provided
Tools the agent exposes — your AI client calls them automatically when it needs them.
- shape_request — Turn a raw request into a buildable one by running the questioning loop: measure readiness, ask the single most valuable question, and repeat as answers come back. Returns the matured request once confidence clears the threshold, or once the turn ceiling is reached - with every unanswered slot written down as an open risk rather than guessed at.
- assess_readiness — Measure how close a request is to being buildable: score each of the eight readiness slots as known, partial, unknown or declined, and return an explainable confidence out of 100 with the points each gap is costing. Deterministic — the score is a weighted sum you can check.
- ask_next_question — Choose the single question that most reduces uncertainty about a request, given what is already known, and report the points that answer would be worth. Returns no question - and names which of the three stop conditions applied - once the request is ready, the turn ceiling is reached, or every slot has been asked once.
- classify_request — Classify a matured request as a bug, a feature or tech debt, report its urgency, and name the downstream flow it should enter. Cue scoring with a published margin: a request that does not commit to a type comes back unclear, and urgency with no cue behind it comes back unstated rather than normal.
- write_use_case — Write an IEEE 830 / UML use case for a feature: UC id, actors, precondition, numbered main flow, alternative flows and acceptance criteria, plus a completeness grade for the input it was given. Every line is derived from what you supplied; actors inferred from prose are reported separately so you can confirm them.
- define_acceptance — Derive Given/When/Then acceptance criteria for a feature, and judge each one: unmet when it can be tested and is simply not delivered yet, unknown when a vague term with no measurable value makes it unjudgeable. Every criterion traces to a supplied input; no threshold is invented.
- split_into_tickets — Split a spec into an epic and a set of ticket drafts, each carrying acceptance criteria taken from its own section, an estimate from the priority heuristic, and dependency hints. Proposal only: this agent creates nothing in any tracker, and every draft is returned requiring confirmation.
- design_process_flow — Turn a written process description into a normalized flow spec - trigger, actions, conditions, ordered steps - and validate it: schema, reference integrity, reachability from the entry node, and cycle detection. The spec is data and is never executed.
- run_full — Run every capability you supplied the input for, in registry order, and return one report. Capabilities whose input is missing are listed as skipped with the fields they need — never run against a fixture and never reported as zero.
- list_capabilities — List what this agent can do: every tool it advertises, what each one is for, and the arguments each one reads. Useful for discovery before the first call.
- plan_inputs — Plan the inputs for a tool: returns the questions to ask, the JSON schema of the arguments, and a ready-to-edit example. Every argument it names is an argument the tool actually reads.
- discover_intent — Understand your goal and co-design the exact input through clarifying questions before anything runs.
- open_form — Use this the moment the user wants to run this agent on their own input. It opens an interactive form that collects the run arguments. Do NOT check credentials, do NOT ask clarifying questions in chat, and do NOT run any analysis before calling this — the form is the entry point for running the agent.
- run_form — Internal: invoked by the open_form form when the user submits. Reads exactly the same arguments as run_full and forwards all of them. Not for direct use.
