@esoteric-logic/praxis-harness 2.17.0 → 3.0.1

package/README.md

@@ -79,6 +79,66 @@ For technical research: `/px-discover` (structured options evaluation before dec
 | `px-context-probe` | Assess context health and recommend action |
 | `px-context-reset` | Reload context from vault without clearing session |
 
+## Prompt Engine
+
+Build and manage system prompts for Claude Projects, Perplexity Spaces, and Claude Code from a single source of truth.
+
+### Quick reference
+
+| Command | What it does |
+|---------|-------------|
+| `/px-prompt my-project` | Create a new project (1 question — describe it) |
+| `/px-prompt --deal irs-masterfile` | New Maximus capture deal (0 questions — Perplexity researches) |
+| `/px-prompt --deal benefeds --type recompete` | New deal with explicit type |
+| `/px-prompt --advance irs-masterfile` | Move deal to next capture phase |
+| `/px-prompt --edit my-project "add NIST 800-53"` | Edit one section, auto-regenerate outputs |
+| `/px-prompt --refresh my-project` | Re-run Perplexity research, diff and update |
+| `/px-prompt --deploy my-project` | Copy outputs to clipboard with platform URLs |
+| `/px-prompt --dashboard` | Project index with budgets and staleness |
+| `/px-prompt --sync` | Recompile all projects, show diffs |
+| `/px-prompt --scan my-project` | Full quality and budget audit |
+
+### Deal lifecycle (Maximus captures)
+
+```
+/px-prompt --deal irs-masterfile          # Create — Perplexity researches from name alone
+/px-prompt --advance irs-masterfile       # Shaping → Mid Capture
+/px-prompt --advance irs-masterfile       # Mid Capture → Pre-Proposal
+/px-prompt --refresh irs-masterfile       # Re-run OSINT before submission
+/px-prompt --advance irs-masterfile       # Pre-Proposal → Pre-Submission
+/px-prompt --deploy irs-masterfile        # Copy to clipboard, paste to platforms
+```
+
+### Deal types
+
+| Type | Flag | When | Emphasis |
+|------|------|------|----------|
+| Recompete | `--type recompete` | Incumbent holds the contract | Defense, ghost matrix, transition |
+| New Start | `--type new-start` | No incumbent, fresh opportunity | Discovery, innovation, OSINT |
+| Task Order | `--type task-order` | Competing on existing IDIQ/BPA | Speed, rates, vehicle history |
+| IDIQ/BPA | `--type idiq` | Winning the vehicle itself | Broad capabilities, teaming |
+
+Auto-detected from research if `--type` not specified.
+
+### Non-Maximus projects
+
+```
+/px-prompt cybersecurity-advisor          # Describe it → inference engine → research → deploy
+/px-prompt elect-azure                    # Work on existing project
+```
+
+Any project — federal, enterprise, personal. The engine infers role, domains, and platforms from your description.
+
+### CLI tools (direct use)
+
+```bash
+node bin/prompt-compile.js --dashboard    # Project dashboard
+node bin/prompt-compile.js --sync         # Recompile all projects
+node bin/prompt-compile.js <project>      # Compile one project
+node bin/prompt-knowledge.js <project>    # Render knowledge pack templates
+node bin/prompt-blocks.js --category domains  # List available blocks
+```
+
 ## Rules
 
 16 rules across universal and scoped categories. Universal rules load every session. Scoped rules load only when matching file patterns are detected.

package/base/hooks/settings-hooks.json

@@ -63,16 +63,16 @@
         "matcher": "",
         "hooks": [
           {
-            "type": "prompt",
-            "prompt": "Run the project test suite. Read CLAUDE.md ## Commands for the test command. If no test command defined, respond {ok:true}. Run tests. If all pass: respond {ok:true}. If tests fail: respond {ok:false, reason:'Tests failing: <summary>'}. Do not fix — only report."
+            "type": "command",
+            "command": "bash ~/.claude/hooks/stop-prompt-gate.sh test-runner"
           },
           {
             "type": "command",
             "command": "bash ~/.claude/hooks/session-data-collect.sh"
           },
           {
-            "type": "prompt",
-            "prompt": "You are completing a session. Perform these vault updates silently — no confirmation needed.\n\n1. Read vault_path from ~/.claude/praxis.config.json. If missing: skip all steps.\n2. Read {vault_path}/.session-staging.json if it exists (structured session data from hooks).\n3. Update {vault_path}/status.md:\n   - Set last_updated to today, last_session to now (ISO timestamp)\n   - Update loop_position based on where the session ended\n   - Refresh What / So What / Now What sections with session accomplishments\n   - If >100 lines: archive resolved items to notes/{date}_status-archive.md\n4. Update {vault_path}/claude-progress.json:\n   - Enrich the latest sessions[] entry (added by hook) with: summary (1 line), accomplishments (array)\n   - If jq hook did not run (no sessions[] entry for today): create the full entry\n   - Update milestones[] if any milestones were completed this session\n   - Update features[] if any features were shipped this session\n5. Write {vault_path}/notes/{YYYY-MM-DD}_session-note.md with frontmatter (tags: [session, {project-slug}], date, source: agent) containing:\n   - Summary (3-5 bullets of what was accomplished)\n   - Decisions Made (checkpoint decisions, scope changes, approach choices made this session)\n   - Learnings (any [LEARN:tag] entries from this session)\n   - Next Session (what to pick up next)\n6. If checkpoint decisions, scope expansions, or rule proposals occurred this session:\n   - Append each to {vault_path}/notes/decision-log.md with date, decision type, context, decision, and rationale\n7. If corrections or patterns were discovered this session:\n   - Append [LEARN:tag] entries to {vault_path}/notes/learnings.md following the existing format\n8. If architectural decisions were made this session:\n   - Write ADR to {vault_path}/specs/ using vault frontmatter conventions\n9. Delete {vault_path}/.session-staging.json if it exists.\n\nKeep all writes concise. Use [[wikilinks]] for internal references. Follow existing YAML frontmatter conventions. If vault_path is missing or vault is inaccessible: skip silently. Do not ask permission — this is automatic housekeeping."
+            "type": "command",
+            "command": "bash ~/.claude/hooks/stop-prompt-gate.sh vault-update"
           }
         ]
       }

package/base/hooks/stop-prompt-gate.sh

@@ -0,0 +1,88 @@
+#!/usr/bin/env bash
+# stop-prompt-gate.sh — Gates Stop prompt hooks behind session activity detection.
+# Only outputs the prompt text if the session had substantive work.
+# Silent exit otherwise — prevents infinite Stop hook loops.
+#
+# Usage in settings-hooks.json:
+#   { "type": "command", "command": "bash ~/.claude/hooks/stop-prompt-gate.sh vault-update" }
+#   { "type": "command", "command": "bash ~/.claude/hooks/stop-prompt-gate.sh test-runner" }
+set -uo pipefail
+
+PROMPT_TYPE="${1:-vault-update}"
+CONFIG_FILE="$HOME/.claude/praxis.config.json"
+
+# ── Detect session activity ──
+HAS_RECENT_COMMITS=false
+HAS_DIRTY_TREE=false
+
+RECENT=$(git --no-pager log --oneline --since="1 hour ago" 2>/dev/null | head -1)
+if [[ -n "$RECENT" ]]; then
+  HAS_RECENT_COMMITS=true
+fi
+
+DIRTY=$(git --no-pager status --porcelain 2>/dev/null | head -1)
+if [[ -n "$DIRTY" ]]; then
+  HAS_DIRTY_TREE=true
+fi
+
+# Check staging file from session-data-collect.sh
+STAGING_DIRTY=false
+if [[ -f "$CONFIG_FILE" ]]; then
+  VAULT_PATH=$(jq -r '.vault_path // empty' "$CONFIG_FILE" 2>/dev/null)
+  STAGING_FILE="${VAULT_PATH:+$VAULT_PATH/.session-staging.json}"
+  if [[ -n "$STAGING_FILE" && -f "$STAGING_FILE" ]]; then
+    STAGING_DIRTY=$(jq -r '.git.dirty // false' "$STAGING_FILE" 2>/dev/null)
+  fi
+fi
+
+# ── Gate: no activity → exit silently ──
+if [[ "$HAS_RECENT_COMMITS" == "false" && "$HAS_DIRTY_TREE" == "false" && "$STAGING_DIRTY" != "true" ]]; then
+  exit 0
+fi
+
+# ── Activity detected — output the requested prompt ──
+case "$PROMPT_TYPE" in
+  test-runner)
+    cat << 'PROMPT'
+Run the project test suite. Read CLAUDE.md ## Commands for the test command. If no test command defined, respond {ok:true}. Run tests. If all pass: respond {ok:true}. If tests fail: respond {ok:false, reason:'Tests failing: <summary>'}. Do not fix — only report.
+PROMPT
+    ;;
+  vault-update)
+    cat << 'PROMPT'
+You are completing a session with substantive work. Perform these vault updates silently — no confirmation needed.
+
+1. Read vault_path from ~/.claude/praxis.config.json. If missing: skip all steps.
+2. Read {vault_path}/.session-staging.json if it exists (structured session data from hooks).
+3. Update {vault_path}/status.md:
+   - Set last_updated to today, last_session to now (ISO timestamp)
+   - Update loop_position based on where the session ended
+   - Refresh What / So What / Now What sections with session accomplishments
+   - If >100 lines: archive resolved items to notes/{date}_status-archive.md
+4. Update {vault_path}/claude-progress.json:
+   - Enrich the latest sessions[] entry (added by hook) with: summary (1 line), accomplishments (array)
+   - If jq hook did not run (no sessions[] entry for today): create the full entry
+   - Update milestones[] if any milestones were completed this session
+   - Update features[] if any features were shipped this session
+5. Write {vault_path}/notes/{YYYY-MM-DD}_session-note.md with frontmatter (tags: [session, {project-slug}], date, source: agent) containing:
+   - Summary (3-5 bullets of what was accomplished)
+   - Decisions Made (checkpoint decisions, scope changes, approach choices made this session)
+   - Learnings (any [LEARN:tag] entries from this session)
+   - Next Session (what to pick up next)
+6. If checkpoint decisions, scope expansions, or rule proposals occurred this session:
+   - Append each to {vault_path}/notes/decision-log.md with date, decision type, context, decision, and rationale
+7. If corrections or patterns were discovered this session:
+   - Append [LEARN:tag] entries to {vault_path}/notes/learnings.md following the existing format
+8. If architectural decisions were made this session:
+   - Write ADR to {vault_path}/specs/ using vault frontmatter conventions
+9. Delete {vault_path}/.session-staging.json if it exists.
+
+Keep all writes concise. Use [[wikilinks]] for internal references. Follow existing YAML frontmatter conventions. If vault_path is missing or vault is inaccessible: skip silently. Do not ask permission — this is automatic housekeeping.
+PROMPT
+    ;;
+  *)
+    echo "Unknown prompt type: $PROMPT_TYPE" >&2
+    exit 0
+    ;;
+esac
+
+exit 0