technomaton
OrganizationEDPA — Evidence-Driven Proportional Allocation. Derive hours from Git evidence. No timesheets.
Categories
Indexed Skills (11)
edpaadd
Create a new backlog item (Initiative / Epic / Feature / Story / Defect / Event / Risk) — V2 local-first. ID allocated from id_counters.yaml, parent hierarchy validated by MCP edpa_item_create, YAML written under .edpa/backlog/, auto-committed. No GitHub calls at create time; PR-derived signals arrive separately via the contribution-sync workflow.
edpaautocalib
Auto-calibrate EDPA CW signal weights using the Monte Carlo + coordinate-descent optimizer (v1.11+). One target file (cw_heuristics.yaml.tmpl), one metric (MAD on a synthetic corpus), two phases (random sample → coordinate descent). Use when: user says "calibrate CW", "auto-calibrate", "optimize heuristics", "recalibrate signals". Synthetic corpus — runnable any time, no ground-truth file required. Re-run after a real PI close once team-confirmed CW corrections are available (see "Re-run with real data" below).
edpaengine
Run EDPA evidence-driven calculation for an iteration. Gathers GitHub delivery evidence (commits, PRs, reviews, comments), computes CW from heuristics, calculates Score and DerivedHours, validates invariants. Use when closing an iteration, computing derived hours, or running "EDPA výpočet". Produces per-person allocation data for the reports skill.
edpareports
Generate EDPA timesheets, reports, and exports. Produces per-person MD/JSON reports, per-item cost allocation, PI summaries, Excel exports, and frozen snapshots. Use when user asks for "reports", "výkazy", "export", "snapshot", or "per-item analysis". Requires edpa-engine results (edpa_results.json) as input.
edpaserver
Start/stop/status of the optional PI planning HTTP server (V2 experimental). Server is a React+Vite frontend backed by an Express+TS API that proxies reads/writes through the EDPA MCP server (single source of truth). Off by default — opt-in via install.sh --with-server. Listens on localhost:3001.
edpasetup
Initialize EDPA V2 governance for a project. Vendors the engine (scripts + schemas + templates) into `.edpa/engine/`, creates `.edpa/config/{edpa.yaml,people.yaml}`, seeds id_counters.yaml, and optionally copies the PR-signal CI workflow + installs git hooks. No GitHub Project provisioning (V1 path removed in 2.0.0).
add
Create a new backlog item (Initiative / Epic / Feature / Story / Defect / Event / Risk) — V2 local-first. ID allocated from id_counters.yaml, parent hierarchy validated by MCP edpa_item_create, YAML written under .edpa/backlog/, auto-committed. No GitHub calls at create time; PR-derived signals arrive separately via the contribution-sync workflow.
autocalib
Auto-calibrate EDPA CW signal weights using the Monte Carlo + coordinate-descent optimizer (v1.11+). One target file (cw_heuristics.yaml.tmpl), one metric (MAD on a synthetic corpus), two phases (random sample → coordinate descent). Use when: user says "calibrate CW", "auto-calibrate", "optimize heuristics", "recalibrate signals". Synthetic corpus — runnable any time, no ground-truth file required. Re-run after a real PI close once team-confirmed CW corrections are available (see "Re-run with real data" below).
engine
Run EDPA evidence-driven calculation for an iteration. Gathers GitHub delivery evidence (commits, PRs, reviews, comments), computes CW from heuristics, calculates Score and DerivedHours, validates invariants. Use when closing an iteration, computing derived hours, or running "EDPA výpočet". Produces per-person allocation data for the reports skill.
reports
Generate EDPA timesheets, reports, and exports. Produces per-person MD/JSON reports, per-item cost allocation, PI summaries, Excel exports, and frozen snapshots. Use when user asks for "reports", "výkazy", "export", "snapshot", or "per-item analysis". Requires /edpa:engine results (edpa_results.json) as input.
setup
Initialize EDPA V2 governance for a project. Vendors the engine (scripts + schemas + templates) into `.edpa/engine/`, creates `.edpa/config/{edpa.yaml,people.yaml}`, seeds id_counters.yaml, and optionally copies the PR-signal CI workflow + installs git hooks. No GitHub Project provisioning (V1 path removed in 2.0.0).
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.