Skip to content
Forgeplan

forgeplan fpf rules

forgeplan fpf rules lists every active rule in the First Principles Framework rule engine, grouped by its recommended action bucket: EXPLORE (generate more hypotheses), INVESTIGATE (gather more evidence for existing hypotheses), or EXPLOIT (commit to a proven path).

The rules encode FPF’s trust calculus in machine-executable form. They’re what fpf check evaluates against individual artifacts and what fpf dashboard aggregates into project-level recommendations.

When to use

Section titled “When to use”
  • Before a sprint — skim the rule set to understand what the engine is looking for.
  • When a rule fires unexpectedly — read the full rule text to understand the trigger.
  • While authoring methodology docs — reference rules by ID.
  • While debugging fpf check output — see the full rule definition behind a match.

When NOT to use

Section titled “When NOT to use”

Usage

Section titled “Usage”
forgeplan fpf rules [OPTIONS]

Options

Section titled “Options”
      --flat     Flat priority-linear table instead of action-grouped tree
      --json     Output full rule dump as JSON
  -h, --help     Print help
  -V, --version  Print version

The default view groups rules by action (EXPLORE / INVESTIGATE / EXPLOIT). --flat switches to a single priority-ordered table when you want to scan top-priority rules regardless of action. --json emits the full rule dump for MCP clients and audit scripts.

Examples

Section titled “Examples”
Terminal window
# Default: action-grouped tree
forgeplan fpf rules


# Flat priority-ordered table
forgeplan fpf rules --flat


# Machine-readable full dump
forgeplan fpf rules --json


# Typical review flow
forgeplan fpf rules
forgeplan fpf check PRD-019
forgeplan fpf dashboard

The three action buckets

Section titled “The three action buckets”

Rules map onto the explore→investigate→exploit axis from FPF Part B:

  • EXPLORE — the artifact or context has too few hypotheses. The rule suggests generating alternatives (e.g. “Fewer than 3 hypotheses on Deep depth — run forgeplan reason”).
  • INVESTIGATE — hypotheses exist but evidence is weak or low-congruence. The rule suggests gathering more (e.g. “R_eff below threshold — attach a measurement EvidencePack”).
  • EXPLOIT — trust is strong enough to commit. The rule clears the artifact for activation or supersession.

Each rule carries a condition (predicate on artifact metadata), an action label, and a human-readable justification tied to an FPF section (e.g. ref: B.3).

How it fits

Section titled “How it fits”

The rule engine was introduced in PRD-041 and extended in PRD-043 (methodology integrity). It reads artifact state from LanceDB and consults the FPF KB for justification text. Output flows into:

  • fpf check <id> — per-artifact view
  • fpf dashboard — aggregated per-context view
  • forgeplan validate / activate gates (for integrity rules)

See also

Section titled “See also”