agent-method 1.5.1 → 1.5.5

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "agent-method",
-  "version": "1.5.1",
-  "description": "CLI tools for the agent-method methodology — registry-driven routing, validation, and project setup for AI-agent-assisted development",
+  "version": "1.5.5",
+  "description": "CLI tools for the wwa methodology — registry-driven routing, validation, and project setup for AI-agent-assisted development",
   "keywords": [
     "ai-agents",
     "prompt-engineering",
@@ -12,9 +12,10 @@
   ],
   "type": "module",
   "license": "MIT",
-  "author": "agent-method contributors",
+  "author": "wwa contributors",
   "bin": {
-    "agent-method": "bin/agent-method.js"
+    "wwa": "bin/wwa.js",
+    "agent-method": "bin/wwa.js"
   },
   "files": [
     "bin/",
@@ -27,13 +28,15 @@
     "node": ">=18.0.0"
   },
   "dependencies": {
+    "@modelcontextprotocol/sdk": "^1.27.1",
     "chalk": "^5.4.0",
+    "chokidar": "^4.0.3",
     "commander": "^12.0.0",
     "inquirer": "^9.0.0",
     "js-yaml": "^4.1.0"
   },
   "repository": {
     "type": "git",
-    "url": "https://github.com/agent-method/agent-method"
+    "url": "https://github.com/anthropics/wwa"
   }
 }

package/templates/README.md

@@ -251,7 +251,7 @@ The methodology works without any tooling. For teams that want additional valida
 ```bash
 npx agent-method                        # zero-install (Node.js 18+)
 npm install -g agent-method             # permanent install
-pip install agent-method-tools          # Python alternative
+pip install wwa-tools          # Python alternative
 ```
 
 ### Developer commands
@@ -271,10 +271,10 @@ pip install agent-method-tools          # Python alternative
 Use friendly names everywhere — all commands accept aliases:
 
 ```bash
-agent-method init code      # software project
-agent-method init context   # analytical/prompt project (e.g. PromptStudy)
-agent-method init data      # data index/querying project (e.g. SysMLv2)
-agent-method init mix       # multi-type project
+wwa init code      # software project
+wwa init context   # analytical/prompt project (e.g. PromptStudy)
+wwa init data      # data index/querying project (e.g. SysMLv2)
+wwa init mix       # multi-type project
 ```
 
 ### Advanced: pipeline subcommands
@@ -283,11 +283,15 @@ For debugging routing logic: `npx agent-method pipeline classify|select|resolve|
 
 ### Dependencies
 
+**Node.js (npx / npm)**:
+- Node.js 18+
+- commander ^12.0, js-yaml ^4.1, inquirer ^9.0, chalk ^5.0
+
+**Python (pip)**:
 - Python 3.9+
-- PyYAML >= 6.0
-- Click >= 8.0
+- PyYAML >= 6.0, Click >= 8.0
 
 ### Future enhancements
 
-- MCP server: `pip install agent-method-tools[mcp]` — exposes pipeline as agent-callable tools
-- Registry watcher: `pip install agent-method-tools[watch]` — proactive validation on file changes
+- MCP server: `pip install wwa-tools[mcp]` — exposes pipeline as agent-callable tools
+- Registry watcher: `pip install wwa-tools[watch]` — proactive validation on file changes

package/templates/entry-points/.cursorrules

@@ -36,7 +36,7 @@ When a file changes, check this table and update dependent files in the same res
 | Project structure | .context/BASE.md (codebase map), this file (if new query types needed) |
 | Intelligence layer file exceeds 300 lines | Restructure into index + components subdirectory (keep active content, archive completed sections) |
 | New domain area | .context/BASE.md, consider new .context/ specialist, this file (if new scoping row) |
-| Session close | SESSION-LOG.md (append micro-entry — workflow, features, cascades, friction, findings) |
+| Session close or high-effort task completion | SESSION-LOG.md (append metrics entry — effort, ambiguity, context level, tokens, time, user response, refinement delta, workflow, features, cascades, friction, findings) |
 
 <!-- INSTRUCTION: Add project-specific cascade rules below the universal ones above. -->
 
@@ -70,7 +70,7 @@ method_version: 1.5
 
 ## CLI tools (optional)
 
-Available via `npx agent-method` (zero-install) or `pip install agent-method-tools`:
+Available via `npx agent-method` (zero-install) or `pip install wwa-tools`:
 
 | When you want to... | Run |
 |---------------------|-----|
@@ -95,7 +95,7 @@ Available via `npx agent-method` (zero-install) or `pip install agent-method-too
 - Surface uncertainty as open questions in STATE.md — never guess silently
 - Keep intelligence layer files under 300 lines — split into index + components subdirectory when exceeded
 - Propose plans and wait for approval — the human controls direction
-- At session close, append a micro-entry to SESSION-LOG.md — never skip, never read previous entries during normal work
+- At session close or after any high-effort task, append a metrics entry to SESSION-LOG.md — include effort level, question ambiguity, context level, estimated tokens, time, user response (accepted/edited/revised/rejected/redirected), and for medium/high effort tasks: revision count, refinement magnitude (none/minor/moderate/major/rework), delta categories, and survival rate. Never skip, never read previous entries during normal work
 
 ## Do not
 

package/templates/entry-points/AGENT.md

@@ -36,7 +36,7 @@ When a file changes, check this table and update dependent files in the same res
 | Project structure | .context/BASE.md (codebase map), this file (if new query types needed) |
 | Intelligence layer file exceeds 300 lines | Restructure into index + components subdirectory (keep active content, archive completed sections) |
 | New domain area | .context/BASE.md, consider new .context/ specialist, this file (if new scoping row) |
-| Session close | SESSION-LOG.md (append micro-entry — workflow, features, cascades, friction, findings) |
+| Session close or high-effort task completion | SESSION-LOG.md (append metrics entry — effort, ambiguity, context level, tokens, time, user response, refinement delta, workflow, features, cascades, friction, findings) |
 
 <!-- INSTRUCTION: Add project-specific cascade rules below the universal ones above. -->
 
@@ -70,7 +70,7 @@ method_version: 1.5
 
 ## CLI tools (optional)
 
-Available via `npx agent-method` (zero-install) or `pip install agent-method-tools`:
+Available via `npx agent-method` (zero-install) or `pip install wwa-tools`:
 
 | When you want to... | Run |
 |---------------------|-----|
@@ -95,7 +95,7 @@ Available via `npx agent-method` (zero-install) or `pip install agent-method-too
 - Surface uncertainty as open questions in STATE.md — never guess silently
 - Keep intelligence layer files under 300 lines — split into index + components subdirectory when exceeded
 - Propose plans and wait for approval — the human controls direction
-- At session close, append a micro-entry to SESSION-LOG.md — never skip, never read previous entries during normal work
+- At session close or after any high-effort task, append a metrics entry to SESSION-LOG.md — include effort level, question ambiguity, context level, estimated tokens, time, user response (accepted/edited/revised/rejected/redirected), and for medium/high effort tasks: revision count, refinement magnitude (none/minor/moderate/major/rework), delta categories, and survival rate. Never skip, never read previous entries during normal work
 
 ## Do not
 

package/templates/entry-points/CLAUDE.md

@@ -36,7 +36,7 @@ When a file changes, check this table and update dependent files in the same res
 | Project structure | .context/BASE.md (codebase map), this file (if new query types needed) |
 | Intelligence layer file exceeds 300 lines | Restructure into index + components subdirectory (keep active content, archive completed sections) |
 | New domain area | .context/BASE.md, consider new .context/ specialist, this file (if new scoping row) |
-| Session close | SESSION-LOG.md (append micro-entry — workflow, features, cascades, friction, findings) |
+| Session close or high-effort task completion | SESSION-LOG.md (append metrics entry — effort, ambiguity, context level, tokens, time, user response, refinement delta, workflow, features, cascades, friction, findings) |
 
 <!-- INSTRUCTION: Add project-specific cascade rules below the universal ones above. -->
 
@@ -70,7 +70,7 @@ method_version: 1.5
 
 ## CLI tools (optional)
 
-Available via `npx agent-method` (zero-install) or `pip install agent-method-tools`:
+Available via `npx agent-method` (zero-install) or `pip install wwa-tools`:
 
 | When you want to... | Run |
 |---------------------|-----|
@@ -95,7 +95,7 @@ Available via `npx agent-method` (zero-install) or `pip install agent-method-too
 - Surface uncertainty as open questions in STATE.md — never guess silently
 - Keep intelligence layer files under 300 lines — split into index + components subdirectory when exceeded
 - Propose plans and wait for approval — the human controls direction
-- At session close, append a micro-entry to SESSION-LOG.md — never skip, never read previous entries during normal work
+- At session close or after any high-effort task, append a metrics entry to SESSION-LOG.md — include effort level, question ambiguity, context level, estimated tokens, time, user response (accepted/edited/revised/rejected/redirected), and for medium/high effort tasks: revision count, refinement magnitude (none/minor/moderate/major/rework), delta categories, and survival rate. Never skip, never read previous entries during normal work
 
 ## Do not
 

package/templates/full/.cursorrules

@@ -41,7 +41,7 @@ When a file changes, check this table and update dependent files in the same res
 | File split, created, deleted, or renamed | .context/REGISTRY.md (file tree, topic index, structural log) |
 | New domain area | .context/BASE.md, consider new .context/ specialist, this file (if new scoping row) |
 | Lifecycle stage change | PROJECT-PROFILE.md (update stage + date), STATE.md (record decision) |
-| Session close | SESSION-LOG.md (append micro-entry — workflow, features, cascades, friction, findings) |
+| Session close or high-effort task completion | SESSION-LOG.md (append metrics entry — effort, ambiguity, context level, tokens, time, user response, refinement delta, workflow, features, cascades, friction, findings) |
 
 <!-- INSTRUCTION: Add project-specific cascade rules below the universal ones above. -->
 
@@ -87,7 +87,7 @@ method_version: 1.5
 
 ## CLI tools (optional)
 
-Available via `npx agent-method` (zero-install) or `pip install agent-method-tools`:
+Available via `npx agent-method` (zero-install) or `pip install wwa-tools`:
 
 | When you want to... | Run |
 |---------------------|-----|
@@ -113,7 +113,7 @@ Available via `npx agent-method` (zero-install) or `pip install agent-method-too
 - Keep intelligence layer files under 300 lines — split into index + components subdirectory when exceeded
 - Propose plans and wait for approval — the human controls direction
 - SUMMARY.md entries follow audit trail format: date, plan, outcome, files, decisions, next
-- At session close, append a micro-entry to SESSION-LOG.md — never skip, never read previous entries during normal work
+- At session close or after any high-effort task, append a metrics entry to SESSION-LOG.md — include effort level, question ambiguity, context level, estimated tokens, time, user response (accepted/edited/revised/rejected/redirected), and for medium/high effort tasks: revision count, refinement magnitude (none/minor/moderate/major/rework), delta categories, and survival rate. Never skip, never read previous entries during normal work
 
 ## Do not
 

package/templates/full/AGENT.md

@@ -41,7 +41,7 @@ When a file changes, check this table and update dependent files in the same res
 | File split, created, deleted, or renamed | .context/REGISTRY.md (file tree, topic index, structural log) |
 | New domain area | .context/BASE.md, consider new .context/ specialist, this file (if new scoping row) |
 | Lifecycle stage change | PROJECT-PROFILE.md (update stage + date), STATE.md (record decision) |
-| Session close | SESSION-LOG.md (append micro-entry — workflow, features, cascades, friction, findings) |
+| Session close or high-effort task completion | SESSION-LOG.md (append metrics entry — effort, ambiguity, context level, tokens, time, user response, refinement delta, workflow, features, cascades, friction, findings) |
 
 <!-- INSTRUCTION: Add project-specific cascade rules below the universal ones above. -->
 
@@ -87,7 +87,7 @@ method_version: 1.5
 
 ## CLI tools (optional)
 
-Available via `npx agent-method` (zero-install) or `pip install agent-method-tools`:
+Available via `npx agent-method` (zero-install) or `pip install wwa-tools`:
 
 | When you want to... | Run |
 |---------------------|-----|
@@ -113,7 +113,7 @@ Available via `npx agent-method` (zero-install) or `pip install agent-method-too
 - Keep intelligence layer files under 300 lines — split into index + components subdirectory when exceeded
 - Propose plans and wait for approval — the human controls direction
 - SUMMARY.md entries follow audit trail format: date, plan, outcome, files, decisions, next
-- At session close, append a micro-entry to SESSION-LOG.md — never skip, never read previous entries during normal work
+- At session close or after any high-effort task, append a metrics entry to SESSION-LOG.md — include effort level, question ambiguity, context level, estimated tokens, time, user response (accepted/edited/revised/rejected/redirected), and for medium/high effort tasks: revision count, refinement magnitude (none/minor/moderate/major/rework), delta categories, and survival rate. Never skip, never read previous entries during normal work
 
 ## Do not
 

package/templates/full/CLAUDE.md

@@ -41,7 +41,7 @@ When a file changes, check this table and update dependent files in the same res
 | File split, created, deleted, or renamed | .context/REGISTRY.md (file tree, topic index, structural log) |
 | New domain area | .context/BASE.md, consider new .context/ specialist, this file (if new scoping row) |
 | Lifecycle stage change | PROJECT-PROFILE.md (update stage + date), STATE.md (record decision) |
-| Session close | SESSION-LOG.md (append micro-entry — workflow, features, cascades, friction, findings) |
+| Session close or high-effort task completion | SESSION-LOG.md (append metrics entry — effort, ambiguity, context level, tokens, time, user response, refinement delta, workflow, features, cascades, friction, findings) |
 
 <!-- INSTRUCTION: Add project-specific cascade rules below the universal ones above. -->
 
@@ -87,7 +87,7 @@ method_version: 1.5
 
 ## CLI tools (optional)
 
-Available via `npx agent-method` (zero-install) or `pip install agent-method-tools`:
+Available via `npx agent-method` (zero-install) or `pip install wwa-tools`:
 
 | When you want to... | Run |
 |---------------------|-----|
@@ -113,7 +113,7 @@ Available via `npx agent-method` (zero-install) or `pip install agent-method-too
 - Keep intelligence layer files under 300 lines — split into index + components subdirectory when exceeded
 - Propose plans and wait for approval — the human controls direction
 - SUMMARY.md entries follow audit trail format: date, plan, outcome, files, decisions, next
-- At session close, append a micro-entry to SESSION-LOG.md — never skip, never read previous entries during normal work
+- At session close or after any high-effort task, append a metrics entry to SESSION-LOG.md — include effort level, question ambiguity, context level, estimated tokens, time, user response (accepted/edited/revised/rejected/redirected), and for medium/high effort tasks: revision count, refinement magnitude (none/minor/moderate/major/rework), delta categories, and survival rate. Never skip, never read previous entries during normal work
 
 ## Do not
 

package/templates/full/SESSION-LOG.md

@@ -1,8 +1,9 @@
 # Session Log
 
-Append-only session observation log for case study data collection. Each session adds a micro-entry at close. This file is read-only during extraction (never during normal work).
+Append-only session observation log for case study data collection. Each session adds a metrics entry at close. High-effort tasks log immediately at task completion. This file is read-only during extraction (never during normal work).
 
-<!-- AGENT INSTRUCTION: At the end of every session, append a new entry below using this format.
+<!-- AGENT INSTRUCTION: At the end of every session — or immediately after any high-effort task —
+     append a new entry below using the format in "Entry format".
      Do NOT read this file during normal work — only append to it.
      Do NOT modify or delete previous entries.
      When this file exceeds 300 lines, archive older entries to session-log/batch-{N}.md -->
@@ -17,9 +18,52 @@ Append-only session observation log for case study data collection. Each session
 | Extension(s) | {code-project / data-exploration / analytical-system / none} |
 | Observation started | {date} |
 
+## Effort classification
+
+| Effort | Description | When to log |
+|--------|-------------|-------------|
+| **low** | Quick answer, single file or no changes, <5 min | At session close |
+| **medium** | Multi-step work, several file changes, 5-30 min | At session close |
+| **high** | Complex multi-file changes, architecture decisions, extensive debugging, 30+ min | Immediately at task completion |
+
+## Assessment scales
+
+**Ambiguity** (agent-assessed — how clear was the user's request?):
+- **low**: Clear, specific request with sufficient context
+- **medium**: Request understood but required interpretation or assumptions
+- **high**: Vague or ambiguous, required significant clarification
+
+**Context level** (agent-assessed — how much project context was loaded?):
+- **very low**: No project files loaded, answered from general knowledge
+- **low**: Entry point only
+- **medium**: Entry point + STATE.md + 1-2 project files
+- **high**: Entry point + STATE.md + specialist context + multiple project files
+- **very high**: Extensive project context, multiple specialists, cross-file analysis
+
+**User response** (agent-observed — how did the user respond to the result?):
+- **accepted**: User proceeded to next step without changes
+- **edited**: User manually modified the agent's output
+- **revised**: User asked agent to redo or revise the result
+- **rejected**: User said no or declined the result entirely
+- **redirected**: User changed approach or gave new instructions
+
+**Refinement magnitude** (for medium/high effort only — how much changed from first attempt to final?):
+- **none**: Accepted as-is, 0% changed
+- **minor**: Small fixes — typos, naming, formatting (<10% changed)
+- **moderate**: Logic or structural changes, added/removed sections (10–50% changed)
+- **major**: Significant rework of approach or content (50–80% changed)
+- **rework**: Mostly rewritten, original approach abandoned (>80% changed)
+
+**Delta categories** (what kinds of changes were needed — select all that apply):
+- **accuracy**: Factual errors or incorrect implementation
+- **completeness**: Missing parts, incomplete coverage
+- **approach**: Wrong method or strategy
+- **scope**: Over-scoped or under-scoped
+- **style**: Formatting, naming, conventions
+
 ## Observation checklist
 
-At session close, reflect on these before writing the micro-entry:
+At session close (or high-effort task completion), reflect on these before writing the entry:
 1. Which workflow did this session follow?
 2. Which query types were encountered?
 3. Which features visibly activated? (context loading, cascade, decision recording, scoping)
@@ -27,15 +71,32 @@ At session close, reflect on these before writing the micro-entry:
 5. Were any decisions deferred instead of recorded immediately?
 6. Was there friction with any methodology rule?
 7. Any degradation signals? (HAI-05: cascade misses, instruction loss, shallow context)
+8. How much effort did this task require? (low / medium / high)
+9. How ambiguous was the user's request? (low / medium / high)
+10. How much project context was loaded? (very low / low / medium / high / very high)
+11. Approximate token usage and time spent?
+12. How did the user respond to the result? (accepted / edited / revised / rejected / redirected)
+13. For medium/high effort: how many revision cycles before acceptance?
+14. What magnitude of change between first attempt and final result? (none / minor / moderate / major / rework)
+15. What categories of refinement were needed? (accuracy / completeness / approach / scope / style)
 
-## Session entries
+## Entry format
 
-<!-- Append new entries below. Format:
+<!-- Append new entries below. Format: -->
+<!--
 ### S{N} — {YYYY-MM-DD} — {brief title}
 Model: {model} | Profile: {profile} | Workflow: {WF-XX}
+Effort: {low / medium / high} | Ambiguity: {low / medium / high} | Context: {very low / low / medium / high / very high}
+Tokens: ~{N}k | Time: ~{N} min
 Queries: {query types encountered}
 Features: {feature IDs activated}
 Cascades: {triggered}/{expected} | Decisions: {count}
+Response: {accepted / edited / revised / rejected / redirected}
+Revisions: {0 | count of revision cycles} | Magnitude: {none / minor / moderate / major / rework}
+Delta: {n/a | categories: accuracy, completeness, approach, scope, style} | Survival: ~{N}%
+Delta notes: {n/a | brief description of what changed between first attempt and final}
 Friction: {none | brief description}
 Finding: {none | observation with methodology implication}
 -->
+
+## Session entries

package/templates/starter/.cursorrules

@@ -39,7 +39,7 @@ When a file changes, check this table and update dependent files in the same res
 | Intelligence layer file exceeds 300 lines | Restructure into index + components subdirectory (keep active content, archive completed sections) |
 | New domain area | .context/BASE.md, consider new .context/ specialist, this file (if new scoping row) |
 | Lifecycle stage change | PROJECT-PROFILE.md (update stage + date), STATE.md (record decision) |
-| Session close | SESSION-LOG.md (append micro-entry — workflow, features, cascades, friction, findings) |
+| Session close or high-effort task completion | SESSION-LOG.md (append metrics entry — effort, ambiguity, context level, tokens, time, user response, refinement delta, workflow, features, cascades, friction, findings) |
 
 <!-- INSTRUCTION: Add project-specific cascade rules below the universal ones above. -->
 
@@ -74,7 +74,7 @@ method_version: 1.5
 
 ## CLI tools (optional)
 
-Available via `npx agent-method` (zero-install) or `pip install agent-method-tools`:
+Available via `npx agent-method` (zero-install) or `pip install wwa-tools`:
 
 | When you want to... | Run |
 |---------------------|-----|
@@ -99,7 +99,7 @@ Available via `npx agent-method` (zero-install) or `pip install agent-method-too
 - Surface uncertainty as open questions in STATE.md — never guess silently
 - Keep intelligence layer files under 300 lines — split into index + components subdirectory when exceeded
 - Propose plans and wait for approval — the human controls direction
-- At session close, append a micro-entry to SESSION-LOG.md — never skip, never read previous entries during normal work
+- At session close or after any high-effort task, append a metrics entry to SESSION-LOG.md — include effort level, question ambiguity, context level, estimated tokens, time, user response (accepted/edited/revised/rejected/redirected), and for medium/high effort tasks: revision count, refinement magnitude (none/minor/moderate/major/rework), delta categories, and survival rate. Never skip, never read previous entries during normal work
 
 ## Do not
 

package/templates/starter/AGENT.md

@@ -39,7 +39,7 @@ When a file changes, check this table and update dependent files in the same res
 | Intelligence layer file exceeds 300 lines | Restructure into index + components subdirectory (keep active content, archive completed sections) |
 | New domain area | .context/BASE.md, consider new .context/ specialist, this file (if new scoping row) |
 | Lifecycle stage change | PROJECT-PROFILE.md (update stage + date), STATE.md (record decision) |
-| Session close | SESSION-LOG.md (append micro-entry — workflow, features, cascades, friction, findings) |
+| Session close or high-effort task completion | SESSION-LOG.md (append metrics entry — effort, ambiguity, context level, tokens, time, user response, refinement delta, workflow, features, cascades, friction, findings) |
 
 <!-- INSTRUCTION: Add project-specific cascade rules below the universal ones above. -->
 
@@ -74,7 +74,7 @@ method_version: 1.5
 
 ## CLI tools (optional)
 
-Available via `npx agent-method` (zero-install) or `pip install agent-method-tools`:
+Available via `npx agent-method` (zero-install) or `pip install wwa-tools`:
 
 | When you want to... | Run |
 |---------------------|-----|
@@ -99,7 +99,7 @@ Available via `npx agent-method` (zero-install) or `pip install agent-method-too
 - Surface uncertainty as open questions in STATE.md — never guess silently
 - Keep intelligence layer files under 300 lines — split into index + components subdirectory when exceeded
 - Propose plans and wait for approval — the human controls direction
-- At session close, append a micro-entry to SESSION-LOG.md — never skip, never read previous entries during normal work
+- At session close or after any high-effort task, append a metrics entry to SESSION-LOG.md — include effort level, question ambiguity, context level, estimated tokens, time, user response (accepted/edited/revised/rejected/redirected), and for medium/high effort tasks: revision count, refinement magnitude (none/minor/moderate/major/rework), delta categories, and survival rate. Never skip, never read previous entries during normal work
 
 ## Do not
 

package/templates/starter/CLAUDE.md

@@ -39,7 +39,7 @@ When a file changes, check this table and update dependent files in the same res
 | Intelligence layer file exceeds 300 lines | Restructure into index + components subdirectory (keep active content, archive completed sections) |
 | New domain area | .context/BASE.md, consider new .context/ specialist, this file (if new scoping row) |
 | Lifecycle stage change | PROJECT-PROFILE.md (update stage + date), STATE.md (record decision) |
-| Session close | SESSION-LOG.md (append micro-entry — workflow, features, cascades, friction, findings) |
+| Session close or high-effort task completion | SESSION-LOG.md (append metrics entry — effort, ambiguity, context level, tokens, time, user response, refinement delta, workflow, features, cascades, friction, findings) |
 
 <!-- INSTRUCTION: Add project-specific cascade rules below the universal ones above. -->
 
@@ -74,7 +74,7 @@ method_version: 1.5
 
 ## CLI tools (optional)
 
-Available via `npx agent-method` (zero-install) or `pip install agent-method-tools`:
+Available via `npx agent-method` (zero-install) or `pip install wwa-tools`:
 
 | When you want to... | Run |
 |---------------------|-----|
@@ -99,7 +99,7 @@ Available via `npx agent-method` (zero-install) or `pip install agent-method-too
 - Surface uncertainty as open questions in STATE.md — never guess silently
 - Keep intelligence layer files under 300 lines — split into index + components subdirectory when exceeded
 - Propose plans and wait for approval — the human controls direction
-- At session close, append a micro-entry to SESSION-LOG.md — never skip, never read previous entries during normal work
+- At session close or after any high-effort task, append a metrics entry to SESSION-LOG.md — include effort level, question ambiguity, context level, estimated tokens, time, user response (accepted/edited/revised/rejected/redirected), and for medium/high effort tasks: revision count, refinement magnitude (none/minor/moderate/major/rework), delta categories, and survival rate. Never skip, never read previous entries during normal work
 
 ## Do not
 

package/templates/starter/SESSION-LOG.md

@@ -1,8 +1,9 @@
 # Session Log
 
-Append-only session observation log for case study data collection. Each session adds a micro-entry at close. This file is read-only during extraction (never during normal work).
+Append-only session observation log for case study data collection. Each session adds a metrics entry at close. High-effort tasks log immediately at task completion. This file is read-only during extraction (never during normal work).
 
-<!-- AGENT INSTRUCTION: At the end of every session, append a new entry below using this format.
+<!-- AGENT INSTRUCTION: At the end of every session — or immediately after any high-effort task —
+     append a new entry below using the format in "Entry format".
      Do NOT read this file during normal work — only append to it.
      Do NOT modify or delete previous entries.
      When this file exceeds 300 lines, archive older entries to session-log/batch-{N}.md -->
@@ -17,9 +18,52 @@ Append-only session observation log for case study data collection. Each session
 | Extension(s) | {code-project / data-exploration / analytical-system / none} |
 | Observation started | {date} |
 
+## Effort classification
+
+| Effort | Description | When to log |
+|--------|-------------|-------------|
+| **low** | Quick answer, single file or no changes, <5 min | At session close |
+| **medium** | Multi-step work, several file changes, 5-30 min | At session close |
+| **high** | Complex multi-file changes, architecture decisions, extensive debugging, 30+ min | Immediately at task completion |
+
+## Assessment scales
+
+**Ambiguity** (agent-assessed — how clear was the user's request?):
+- **low**: Clear, specific request with sufficient context
+- **medium**: Request understood but required interpretation or assumptions
+- **high**: Vague or ambiguous, required significant clarification
+
+**Context level** (agent-assessed — how much project context was loaded?):
+- **very low**: No project files loaded, answered from general knowledge
+- **low**: Entry point only
+- **medium**: Entry point + STATE.md + 1-2 project files
+- **high**: Entry point + STATE.md + specialist context + multiple project files
+- **very high**: Extensive project context, multiple specialists, cross-file analysis
+
+**User response** (agent-observed — how did the user respond to the result?):
+- **accepted**: User proceeded to next step without changes
+- **edited**: User manually modified the agent's output
+- **revised**: User asked agent to redo or revise the result
+- **rejected**: User said no or declined the result entirely
+- **redirected**: User changed approach or gave new instructions
+
+**Refinement magnitude** (for medium/high effort only — how much changed from first attempt to final?):
+- **none**: Accepted as-is, 0% changed
+- **minor**: Small fixes — typos, naming, formatting (<10% changed)
+- **moderate**: Logic or structural changes, added/removed sections (10–50% changed)
+- **major**: Significant rework of approach or content (50–80% changed)
+- **rework**: Mostly rewritten, original approach abandoned (>80% changed)
+
+**Delta categories** (what kinds of changes were needed — select all that apply):
+- **accuracy**: Factual errors or incorrect implementation
+- **completeness**: Missing parts, incomplete coverage
+- **approach**: Wrong method or strategy
+- **scope**: Over-scoped or under-scoped
+- **style**: Formatting, naming, conventions
+
 ## Observation checklist
 
-At session close, reflect on these before writing the micro-entry:
+At session close (or high-effort task completion), reflect on these before writing the entry:
 1. Which workflow did this session follow?
 2. Which query types were encountered?
 3. Which features visibly activated? (context loading, cascade, decision recording, scoping)
@@ -27,15 +71,32 @@ At session close, reflect on these before writing the micro-entry:
 5. Were any decisions deferred instead of recorded immediately?
 6. Was there friction with any methodology rule?
 7. Any degradation signals? (HAI-05: cascade misses, instruction loss, shallow context)
+8. How much effort did this task require? (low / medium / high)
+9. How ambiguous was the user's request? (low / medium / high)
+10. How much project context was loaded? (very low / low / medium / high / very high)
+11. Approximate token usage and time spent?
+12. How did the user respond to the result? (accepted / edited / revised / rejected / redirected)
+13. For medium/high effort: how many revision cycles before acceptance?
+14. What magnitude of change between first attempt and final result? (none / minor / moderate / major / rework)
+15. What categories of refinement were needed? (accuracy / completeness / approach / scope / style)
 
-## Session entries
+## Entry format
 
-<!-- Append new entries below. Format:
+<!-- Append new entries below. Format: -->
+<!--
 ### S{N} — {YYYY-MM-DD} — {brief title}
 Model: {model} | Profile: {profile} | Workflow: {WF-XX}
+Effort: {low / medium / high} | Ambiguity: {low / medium / high} | Context: {very low / low / medium / high / very high}
+Tokens: ~{N}k | Time: ~{N} min
 Queries: {query types encountered}
 Features: {feature IDs activated}
 Cascades: {triggered}/{expected} | Decisions: {count}
+Response: {accepted / edited / revised / rejected / redirected}
+Revisions: {0 | count of revision cycles} | Magnitude: {none / minor / moderate / major / rework}
+Delta: {n/a | categories: accuracy, completeness, approach, scope, style} | Survival: ~{N}%
+Delta notes: {n/a | brief description of what changed between first attempt and final}
 Friction: {none | brief description}
 Finding: {none | observation with methodology implication}
 -->
+
+## Session entries