@garethdaine/agentops 0.9.0

package/commands/code-analysis.md

@@ -0,0 +1,39 @@
+---
+name: code-analysis
+description: Run a structured code analysis on the current project or specified path
+---
+
+Run a structured code analysis using the `code-critic` agent.
+
+## Process
+
+1. **Determine scope:** If `$ARGUMENTS` specifies a path or glob pattern, use that. Otherwise, analyze the full project working directory.
+2. **Snapshot:** Use Glob and Grep to build a quick inventory of the codebase:
+   - File count and types (by extension)
+   - Tech stack detection (package.json, composer.json, Cargo.toml, go.mod, requirements.txt, Gemfile, etc.)
+   - Entry points and key structural files
+3. **Plan analysis tasks:** Based on the detected stack, select relevant analysis passes from:
+   - **Filesystem manifest:** directory structure, file distribution, naming conventions
+   - **Dependency audit:** outdated/unused/duplicate dependencies, version constraints
+   - **Architecture patterns:** separation of concerns, layering, module boundaries
+   - **Code quality:** readability, naming, DRY, SOLID violations, complexity hotspots
+   - **Performance risks:** N+1 queries, unnecessary allocations, missing indexes, blocking I/O
+   - **Test coverage map:** test file to source file mapping, coverage gaps, assertion quality
+   - **Risk hotspots:** files with high churn + high complexity, large files, deeply nested logic
+   - **Security surface:** credential exposure, injection vectors, auth gaps (delegate to `security-reviewer` agent if available)
+4. **Execute:** Dispatch the `code-critic` agent (and optionally `security-reviewer` agent in parallel) to perform each analysis pass. Each agent should reference exact file paths and line numbers.
+5. **Report:** Compile findings into a structured report and write it to `docs/discovery/code-analysis/{project-name}.md` with:
+   - Summary stats (files, lines, stack, dependencies)
+   - Findings by category, ordered by severity (critical > high > medium > low)
+   - Concrete recommendations with file:line references
+   - A risk score (0-100) based on finding density and severity
+
+## IMPORTANT: This is a read-only, reporting tool.
+
+- Do NOT fix, patch, or modify any code during or after analysis.
+- Do NOT edit any files other than the report output in `docs/discovery/code-analysis/`.
+- Present the report summary to the user and STOP.
+- Wait for explicit user instructions before taking any action on findings.
+- If the user asks you to fix findings, treat that as a separate task requiring its own STAR plan.
+
+Arguments: $ARGUMENTS

package/commands/code-field.md

@@ -0,0 +1,22 @@
+---
+name: code-field
+description: Apply the Code Field methodology (decompose, solve, verify, synthesize, reflect)
+---
+
+Apply the **Code Field** methodology to the current task or technical answer:
+
+1. **DECOMPOSE** — Sub-problems: parts, dependencies, order.
+2. **SOLVE** — Each part with confidence 0.0–1.0 (0.9+ proceed; 0.7–0.8 extra test; 0.5–0.6 verify first; <0.5 research/clarify).
+3. **VERIFY** — Logic, facts, completeness, bias.
+4. **SYNTHESIZE** — Weighted confidence; flag anything <0.7.
+5. **REFLECT** — If <0.8 overall: weakness, research/ask, retry.
+
+For complex solutions, end with:
+
+```
+✅ Answer: [clear solution or recommendation]
+📊 Confidence: [0.0–1.0] ([brief justification])
+⚠️ Caveats: [assumptions, limitations, conditions]
+```
+
+Arguments: $ARGUMENTS

package/commands/compliance-check.md

@@ -0,0 +1,34 @@
+---
+name: compliance-check
+description: Run AgentOps compliance checks manually
+---
+
+Run all AgentOps compliance gates and report their status:
+
+## Checks
+
+### 1. Plan Gate
+- Count modified files tracked in `.agentops/modified-files.txt`
+- If 3+ files modified, verify `tasks/todo.md` exists and is non-empty
+- Report: PASS/FAIL with file count
+
+### 2. Verification Gate
+- Read `tasks/todo.md` and count unchecked items (`- [ ]`)
+- Report: PASS/FAIL with count of remaining items
+
+### 3. Test Gate
+- Check if `.agentops/tests-ran` exists (populated by ci-guard hook)
+- If 3+ files modified and no tests run, report FAIL
+- Report: PASS/FAIL with last test timestamp if available
+
+### 4. Lessons Check
+- Check if `tasks/lessons.md` exists
+- Count total lessons
+- Report: INFO with lesson count
+
+Present results in a table format:
+
+| Gate | Status | Details |
+|------|--------|---------|
+
+Arguments: $ARGUMENTS

package/commands/configure.md

@@ -0,0 +1,178 @@
+---
+name: configure
+description: Unified configuration interface for all AgentOps plugin settings
+---
+
+You are the AgentOps configuration assistant. You provide a single entry point for viewing and modifying all plugin settings.
+
+**Important:** The plugin root is wherever this command file lives — use the `CLAUDE_PLUGIN_ROOT` environment variable or detect the plugin directory by finding `hooks/feature-flags.sh` relative to this command. The configuration file is always at `{plugin_root}/.agentops/flags.json`. If it doesn't exist, create it with defaults.
+
+Arguments: $ARGUMENTS
+
+---
+
+## No Arguments — Show Full Configuration
+
+If no arguments provided, read `.agentops/flags.json` and `.agentops/budget.json` (both relative to the project root), then display the complete configuration:
+
+```
+## AgentOps Configuration
+
+### Enforcement
+| Setting | Value |
+|---------|-------|
+| Enforcement mode | advisory / blocking |
+| Autonomy level | guided / supervised / autonomous |
+
+### Security Hooks
+| Flag | Status |
+|------|--------|
+| command_validation_enabled | on/off |
+| path_validation_enabled | on/off |
+| env_validation_enabled | on/off |
+| injection_scan_enabled | on/off |
+| content_trust_enabled | on/off |
+| exfiltration_detection_enabled | on/off |
+| credential_redaction_enabled | on/off |
+| llm_content_firewall_enabled | on/off |
+
+### Workflow Automation
+| Flag | Status |
+|------|--------|
+| star_preamble_enabled | on/off |
+| code_field_rules_enabled | on/off |
+| plan_gate_enabled | on/off |
+| verification_gate_enabled | on/off |
+| test_gate_enabled | on/off |
+| lessons_enabled | on/off |
+| auto_plan_enabled | on/off |
+| auto_test_enabled | on/off |
+| auto_lesson_enabled | on/off |
+| auto_verify_enabled | on/off |
+| auto_evolve_enabled | on/off |
+| auto_delegate_enabled | on/off |
+
+### Enterprise Features
+| Flag | Status |
+|------|--------|
+| enterprise_scaffold | on/off |
+| ai_workflows | on/off |
+| unified_review | on/off |
+| architecture_guardrails | on/off |
+| delivery_lifecycle | on/off |
+| team_governance | on/off |
+| client_comms | on/off |
+
+### Budget
+| Setting | Value |
+|---------|-------|
+| Budget | $X.XX |
+| Spent | $X.XX |
+| Remaining | $X.XX |
+| Warning threshold | X% |
+
+To change a setting: `/agentops:configure <setting> <value>`
+```
+
+---
+
+## With Arguments — Update Settings
+
+Parse the arguments as `<setting> <value>` and update accordingly.
+
+### Supported Settings
+
+**Modes:**
+- `enforcement advisory` or `enforcement blocking` — set enforcement mode
+- `autonomy guided`, `autonomy supervised`, or `autonomy autonomous` — set autonomy level
+
+**Budget:**
+- `budget <amount>` — set session budget in USD (e.g., `budget 10`)
+- `budget-warn <percent>` — set warning threshold percentage (e.g., `budget-warn 90`)
+
+**Feature flags (any flag name):**
+- `<flag_name> on` or `<flag_name> true` — enable a flag
+- `<flag_name> off` or `<flag_name> false` — disable a flag
+
+**Presets (bulk operations):**
+- `preset minimal` — disable all automation hooks, keep security hooks on
+- `preset standard` — all defaults (security on, automation on, enterprise on)
+- `preset security-only` — security hooks on, everything else off
+- `preset enterprise` — all enterprise features on, autonomy guided
+- `preset autonomous` — all enterprise features on, autonomy autonomous
+
+### Preset Definitions
+
+**minimal:**
+```json
+{
+  "auto_plan_enabled": false, "auto_test_enabled": false,
+  "auto_lesson_enabled": false, "auto_verify_enabled": false,
+  "auto_evolve_enabled": false, "auto_delegate_enabled": false,
+  "plan_gate_enabled": false, "verification_gate_enabled": false,
+  "test_gate_enabled": false, "star_preamble_enabled": false,
+  "code_field_rules_enabled": false
+}
+```
+
+**security-only:**
+All automation and enterprise flags set to `false`. All security flags remain `true`.
+
+**enterprise:**
+All flags `true`, `enforcement_mode: "advisory"`, `autonomy_level: "guided"`.
+
+**autonomous:**
+All flags `true`, `enforcement_mode: "advisory"`, `autonomy_level: "autonomous"`.
+
+---
+
+## Update Process
+
+1. Read the current `.agentops/flags.json` (create with defaults if missing)
+2. Apply the requested change
+3. Write back the updated JSON
+4. Display confirmation: "Updated `<setting>` to `<value>`"
+5. Show the affected section of the configuration table
+
+## Defaults (used when creating flags.json)
+
+```json
+{
+  "command_validation_enabled": true,
+  "path_validation_enabled": true,
+  "env_validation_enabled": true,
+  "injection_scan_enabled": true,
+  "content_trust_enabled": true,
+  "exfiltration_detection_enabled": true,
+  "credential_redaction_enabled": true,
+  "llm_content_firewall_enabled": true,
+  "plan_gate_enabled": true,
+  "verification_gate_enabled": true,
+  "test_gate_enabled": true,
+  "star_preamble_enabled": true,
+  "code_field_rules_enabled": true,
+  "lessons_enabled": true,
+  "auto_plan_enabled": true,
+  "auto_test_enabled": true,
+  "auto_lesson_enabled": true,
+  "auto_verify_enabled": true,
+  "auto_evolve_enabled": true,
+  "auto_delegate_enabled": true,
+  "enforcement_mode": "advisory",
+  "enterprise_scaffold": true,
+  "ai_workflows": true,
+  "unified_review": true,
+  "architecture_guardrails": true,
+  "delivery_lifecycle": true,
+  "team_governance": true,
+  "client_comms": true,
+  "autonomy_level": "guided"
+}
+```
+
+## Error Handling
+
+- If an unrecognised setting name is provided, suggest the closest match (fuzzy)
+- If an invalid value is provided (e.g., `autonomy fast`), show allowed values
+- If `.agentops/` directory doesn't exist, create it
+- Never crash — always provide helpful feedback

package/commands/cost-report.md

@@ -0,0 +1,17 @@
+---
+name: cost-report
+description: Display current session cost tracking and budget status
+---
+
+Read `.agentops/budget.json` and display the current budget status:
+
+- **Budget:** Total allocated USD
+- **Spent:** Amount used so far
+- **Remaining:** Budget minus spent
+- **Warning threshold:** Percentage at which warnings trigger
+- **Status:** OK / WARNING / EXCEEDED
+- **Session started:** Timestamp
+
+If `.agentops/budget.json` doesn't exist, report that budget tracking is not initialized and suggest starting a session with `AGENTOPS_BUDGET_USD` set.
+
+Arguments: $ARGUMENTS

package/commands/enterprise/adr.md

@@ -0,0 +1,78 @@
+---
+name: adr
+description: Generate Architecture Decision Records capturing the why behind technical decisions
+---
+
+You are an Architecture Decision Record (ADR) assistant. You help engineers document technical decisions in a structured, searchable format.
+
+**Before starting, check the feature flag:**
+Run: `source hooks/feature-flags.sh && agentops_enterprise_enabled "architecture_guardrails"` — if disabled, inform the user and stop.
+
+The decision to document: $ARGUMENTS
+
+If no arguments provided, ask: "What technical decision would you like to document as an ADR?"
+
+---
+
+## Step 1: Determine ADR Number
+
+Check `docs/adr/` in the project for existing ADRs. The new ADR number is the next sequential number. If no `docs/adr/` directory exists, create it and start at 0001.
+
+## Step 2: Gather Context
+
+Ask the user:
+1. **What is the decision?** (one sentence)
+2. **What problem does this solve?** (context)
+3. **What alternatives were considered?** (at least 2)
+4. **Why was this option chosen?** (rationale)
+
+If the user provides all context upfront, skip questions and proceed directly.
+
+## Step 3: Generate ADR
+
+Write the ADR to `docs/adr/NNNN-title-in-kebab-case.md`:
+
+```markdown
+# ADR-NNNN: [Title]
+
+## Status
+
+Accepted
+
+## Date
+
+[today's date]
+
+## Context
+
+[What is the issue that motivates this decision? What forces are at play?]
+
+## Decision
+
+[What is the change that we're proposing and/or doing?]
+
+## Consequences
+
+### Positive
+- [benefit]
+
+### Negative
+- [trade-off]
+
+### Neutral
+- [observation]
+
+## Alternatives Considered
+
+### [Alternative 1]
+**Rejected because:** [reason]
+
+### [Alternative 2]
+**Rejected because:** [reason]
+```
+
+## Step 4: Update Index
+
+If a `docs/adr/README.md` or `docs/adr/index.md` exists, add the new ADR to the list. If not, create one with all existing ADRs listed.
+
+Report: "ADR-NNNN created at docs/adr/NNNN-title.md"