@intentsolutionsio/skill-creator 5.0.0 → 5.0.6

package/skills/agent-creator/SKILL.md

@@ -1,20 +1,33 @@
 ---
 name: agent-creator
-description: |
-  Create production-grade agent .md files aligned with the Anthropic 2026 spec (16-field schema).
-  Also validates existing agents against the marketplace compliance rules. Use when building custom
-  subagents, reviewing agent quality, or creating parallel agent architectures for orchestrator skills.
-  Trigger with "/agent-creator", "create an agent", "build a subagent", or "validate my agent".
-  Make sure to use this skill whenever creating agents/*.md files for plugins or standalone use.
-allowed-tools: "Read,Write,Edit,Glob,Grep,Bash(python:*),AskUserQuestion"
+description: 'Create production-grade agent .md files aligned with the Anthropic 2026
+  spec (16-field schema).
+
+  Also validates existing agents against the marketplace compliance rules. Use when
+  building custom
+
+  subagents, reviewing agent quality, or creating parallel agent architectures for
+  orchestrator skills.
+
+  Trigger with "/agent-creator", "create an agent", "build a subagent", or "validate
+  my agent".
+
+  Make sure to use this skill whenever creating agents/*.md files for plugins or standalone
+  use.
+
+  '
+allowed-tools: Read,Write,Edit,Glob,Grep,Bash(python:*),AskUserQuestion
 version: 1.0.0
 author: Jeremy Longshore <jeremy@intentsolutions.io>
 license: MIT
-compatible-with: claude-code, codex, openclaw
-tags: [agent-creation, validation, meta-tooling, subagents]
+tags:
+- agent-creation
+- validation
+- meta-tooling
+- subagents
 model: inherit
+compatibility: Designed for Claude Code, also compatible with Codex and OpenClaw
 ---
-
 # Agent Creator
 
 Creates spec-compliant agent .md files following the Anthropic 2026 16-field schema. Supports
@@ -45,6 +58,7 @@ that drives the subagent — it does NOT receive the full Claude Code system pro
 ### Mode Detection
 
 Determine user intent from their prompt:
+
 - **Create mode**: "create an agent", "build a subagent", "new agent" -> Step 1
 - **Validate mode**: "validate agent", "check agent", "grade agent" -> Validation Workflow
 
@@ -53,21 +67,25 @@ Determine user intent from their prompt:
 Ask the user with AskUserQuestion:
 
 **Agent Identity:**
+
 - Name (kebab-case, 1-64 chars, e.g., `risk-assessor`, `clause-analyzer`)
 - Specialty description (20-200 chars — shown in agent selection UI)
 
 **Execution Context:**
+
 - Plugin agent (`plugins/*/agents/`) or standalone (`~/.claude/agents/`)?
 - Will it be spawned by an orchestrator skill via `Task` tool?
 - Does it need to preload specific skills? (`skills: [skill-name]`)
 
 **Behavioral Controls:**
+
 - Model override? (`sonnet` for speed, `opus` for quality, `inherit` for default)
 - Reasoning effort? (`low` for simple, `medium` default, `high` for complex analysis)
 - Max iterations? (`maxTurns` — how many tool-use loops before stopping)
 - Tools to deny? (`disallowedTools` — denylist approach, opposite of skills)
 
 **Plugin Restrictions (if plugin agent):**
+
 - `hooks` — NOT supported in plugin agents (use plugin-level hooks)
 - `mcpServers` — NOT supported in plugin agents
 - `permissionMode` — standalone only, NOT plugin agents
@@ -78,6 +96,7 @@ Before writing, determine:
 
 **Agent Role Clarity:**
 The agent body must make three things unambiguous:
+
 1. **What it IS responsible for** — its specific domain/methodology
 2. **What it is NOT responsible for** — boundaries with other agents
 3. **How it communicates results** — output format and structure
@@ -99,6 +118,7 @@ All production agents should follow this body structure:
 | `## Examples` | Concrete interaction examples | For complex agents |
 
 **Output Structure Decision:**
+
 - If the agent feeds into an orchestrator: use **JSON output** (machine-parseable)
 - If the agent is user-facing: use **markdown output** (human-readable)
 - If the agent produces both: JSON primary with markdown summary
@@ -113,12 +133,14 @@ Generate the agent .md using the template from
 See [Anthropic Agent Spec](references/anthropic-agent-spec.md) for the full official reference.
 
 Required fields:
+
 ```yaml
 name: {agent-name}         # Lowercase letters and hyphens, unique identifier
 description: "{specialty}"  # When Claude should delegate to this subagent
 ```
 
 Optional fields (include only what's needed):
+
 ```yaml
 tools: "Read, Glob, Grep"  # Allowlist — inherits all tools if omitted
 disallowedTools: "Write"   # Denylist — removed from inherited/specified list
@@ -137,12 +159,14 @@ mcpServers: {}             # Standalone only, NOT plugin agents
 ```
 
 **Tool access:**
+
 - `tools` = allowlist (like skills' `allowed-tools`)
 - `disallowedTools` = denylist (remove specific tools)
 - If both set: disallowed applied first, then tools resolved
 - If neither set: inherits all tools from parent conversation
 
 **Invalid fields (ERROR — never use these):**
+
 - `capabilities` — looks valid but flagged by validator
 - `expertise_level` — invented, not in Anthropic spec
 - `activation_priority` — invented, not in Anthropic spec
@@ -191,6 +215,7 @@ Run validation against the Anthropic 16-field schema:
 | Body under 300 lines | Offload to references if longer (prevents context bloat) |
 
 **Automated validation:**
+
 ```bash
 python3 ${CLAUDE_SKILL_DIR}/../skill-creator/scripts/validate-skill.py --agents-only {plugin-dir}/
 ```
@@ -209,6 +234,7 @@ Test the agent by spawning it via the `Task` tool or the `Agent` tool:
 ### Step 6: Report
 
 Provide a summary:
+
 - Agent name and file path
 - Frontmatter field count (of 14 possible)
 - Body line count
@@ -299,7 +325,7 @@ actionable).
 ## Resources
 
 - [Anthropic Agent Spec](references/anthropic-agent-spec.md) — Official 16-field schema from code.claude.com/docs/en/sub-agents
-- [Agent template](${CLAUDE_SKILL_DIR}/../skill-creator/templates/agent-template.md) — Skeleton with placeholders
-- [Frontmatter spec](${CLAUDE_SKILL_DIR}/../skill-creator/references/frontmatter-spec.md) — Field reference (internal)
-- [Source of truth](${CLAUDE_SKILL_DIR}/../skill-creator/references/source-of-truth.md) — Canonical spec
-- [Validation rules](${CLAUDE_SKILL_DIR}/../skill-creator/references/validation-rules.md) — Agent validation section
+- Agent template — Skeleton with placeholders
+- Frontmatter spec — Field reference (internal)
+- Source of truth — Canonical spec
+- Validation rules — Agent validation section

package/skills/agent-creator/references/anthropic-agent-spec.md

@@ -39,6 +39,7 @@ Total: 16 official fields.
 ## Plugin Agent Restrictions
 
 Plugin agents (`plugins/*/agents/*.md`) do NOT support:
+
 - `hooks` — ignored when loading from plugin
 - `mcpServers` — ignored when loading from plugin
 - `permissionMode` — ignored when loading from plugin

package/skills/skill-creator/SKILL.md

@@ -1,20 +1,31 @@
 ---
 name: skill-creator
-description: |
-  Create production-grade agent skills aligned with the 2026 AgentSkills.io spec and Anthropic
-  best practices (2026). Also validates existing skills against the Intent Solutions 100-point rubric.
+description: 'Create production-grade agent skills aligned with the 2026 AgentSkills.io
+  spec and Anthropic
+
+  best practices (2026). Also validates existing skills against the Intent Solutions
+  100-point rubric.
+
   Use when building, testing, validating, or optimizing Claude Code skills.
-  Trigger with "/skill-creator", "create a skill", "validate my skill", or "check skill quality".
-  Make sure to use this skill whenever creating a new skill, slash command, or agent capability.
-allowed-tools: "Read,Write,Edit,Glob,Grep,Bash(mkdir:*),Bash(chmod:*),Bash(python:*),Bash(claude:*),Task,AskUserQuestion"
+
+  Trigger with "/skill-creator", "create a skill", "validate my skill", or "check
+  skill quality".
+
+  Make sure to use this skill whenever creating a new skill, slash command, or agent
+  capability.
+
+  '
+allowed-tools: Read,Write,Edit,Glob,Grep,Bash(mkdir:*),Bash(chmod:*),Bash(python:*),Bash(claude:*),Task,AskUserQuestion
 version: 5.1.0
 author: Jeremy Longshore <jeremy@intentsolutions.io>
 license: MIT
-compatible-with: claude-code, codex, openclaw
-tags: [skill-creation, validation, meta-tooling]
+tags:
+- skill-creation
+- validation
+- meta-tooling
 model: inherit
+compatibility: Designed for Claude Code, also compatible with Codex and OpenClaw
 ---
-
 # Skill Creator
 
 Creates complete, spec-compliant skill packages following AgentSkills.io and Anthropic standards.
@@ -39,12 +50,14 @@ scratch with full validation, or grade/audit existing skills with actionable fix
 ### Mode Detection
 
 Determine user intent from their prompt:
+
 - **Create mode**: "create a skill", "build a skill", "new skill" -> proceed to Step 1
 - **Validate mode**: "validate", "check", "grade", "score", "audit" -> jump to Validation Workflow
 
 ### Communicating with the User
 
 Pay attention to context cues to understand the user's technical level. Skill creator is used by people across a wide range of familiarity — from first-time coders to senior engineers. In the default case:
+
 - "evaluation" and "benchmark" are borderline but OK
 - For "JSON" and "assertion", check for cues the user knows these terms before using them without explanation
 - Briefly explain terms if in doubt
@@ -56,21 +69,25 @@ If the current conversation already contains a workflow the user wants to captur
 Ask the user with AskUserQuestion:
 
 **Skill Identity:**
+
 - Name (kebab-case, gerund preferred: `processing-pdfs`, `analyzing-data`)
 - Purpose (1-2 sentences: what it does + when to use it)
 
 **Execution Model:**
+
 - User-invocable via `/name`? Or background knowledge only?
 - Accepts arguments? (`$ARGUMENTS` substitution)
 - Needs isolated context? (`context: fork` for subagent execution)
 - Explicit-only invocation? (`disable-model-invocation: true` — prevents auto-activation, requires `/name`)
 
 **Required Tools:**
+
 - Read, Write, Edit, Glob, Grep, WebFetch, WebSearch, Task, AskUserQuestion, Skill
 - Bash must be scoped: `Bash(git:*)`, `Bash(npm:*)`, etc.
 - MCP tools: `ServerName:tool_name`
 
 **Complexity:**
+
 - Simple (SKILL.md only)
 - With scripts (automation code in `scripts/`)
 - With references (documentation in `references/`)
@@ -78,6 +95,7 @@ Ask the user with AskUserQuestion:
 - Full package (all directories)
 
 **Location:**
+
 - Global: `~/.claude/skills/<skill-name>/`
 - Project: `.claude/skills/<skill-name>/`
 
@@ -86,6 +104,7 @@ Ask the user with AskUserQuestion:
 Before writing, determine:
 
 **Degrees of Freedom:**
+
 | Level | When to Use |
 |-------|-------------|
 | High | Creative/open-ended tasks (analysis, writing) |
@@ -95,6 +114,7 @@ Before writing, determine:
 Think of it as **narrow bridge vs open field**: a deployment skill is a narrow bridge (one safe path, guard rails everywhere), while a writing skill is an open field (Claude roams freely within broad boundaries). Match constraint level to the task.
 
 **Workflow Pattern** (see `${CLAUDE_SKILL_DIR}/references/workflows.md`):
+
 - Sequential: fixed steps in order
 - Conditional: branch based on input
 - Wizard: interactive multi-step gathering
@@ -104,6 +124,7 @@ Think of it as **narrow bridge vs open field**: a deployment skill is a narrow b
 - Search-Analyze-Report: explore and summarize
 
 **Output Pattern** (see `${CLAUDE_SKILL_DIR}/references/output-patterns.md`):
+
 - Strict template (exact format)
 - Flexible template (structure with creative content)
 - Examples-driven (input/output pairs)
@@ -128,6 +149,7 @@ mkdir -p {location}/{skill-name}/evals        # for eval-driven development
 For detailed guidance on writing SKILL.md (frontmatter rules, description scoring, body guidelines, string substitutions, DCI syntax), creating supporting files, validation, testing, iteration, description optimization, and final reporting, see [Creation Guide](references/creation-guide.md).
 
 Key rules:
+
 - `version`, `author`, `license`, `tags`, `compatible-with` are TOP-LEVEL fields (not nested under `metadata:`)
 - Scope Bash: `Bash(git:*)` not bare `Bash`
 - Keep under 500 lines; offload to `references/` if longer
@@ -244,15 +266,18 @@ Output:
 ## Resources
 
 **References:** `${CLAUDE_SKILL_DIR}/references/`
+
 - `creation-guide.md` — Detailed Steps 4-10 and Validation Workflow (V1-V5)
 - `source-of-truth.md` — Canonical spec ([AgentSkills.io](https://agentskills.io/specification), [Anthropic docs](https://code.claude.com/docs/en/skills), [Lee Han Chung deep dive](https://leehanchung.github.io/blogs/2025/10/26/claude-skills-deep-dive/)) | `frontmatter-spec.md` — Field reference | `validation-rules.md` — 100-point rubric
 - `workflows.md` — Workflow patterns | `output-patterns.md` — Output formats | `schemas.md` — JSON schemas (evals, grading, benchmarks)
 - `anthropic-comparison.md` — Gap analysis | `advanced-eval-workflow.md` — Eval, iteration, optimization, platform notes
 
 **Agents** (read when spawning subagents): `${CLAUDE_SKILL_DIR}/agents/`
+
 - `grader.md` — Assertion evaluation | `comparator.md` — Blind A/B comparison | `analyzer.md` — Benchmark analysis
 
 **Scripts:** `${CLAUDE_SKILL_DIR}/scripts/`
+
 - `validate-skill.py` — 100-point rubric grading | `quick_validate.py` — Lightweight validation
 - `aggregate_benchmark.py` — Benchmark stats | `run_eval.py` — Trigger accuracy testing
 - `run_loop.py` — Description optimization loop | `improve_description.py` — LLM-powered rewriting

package/skills/skill-creator/agents/analyzer.md

@@ -1,8 +1,35 @@
 ---
 name: analyzer
 description: Analyze blind comparison results to understand why the winner won and generate improvement suggestions
+tools:
+- Read
+- Write
+- Edit
+- Bash
+- Glob
+- Grep
+- WebFetch
+- WebSearch
+- Task
+- TodoWrite
+model: sonnet
+color: orange
+version: 1.0.0
+author: Jeremy Longshore <jeremy@intentsolutions.io>
+tags:
+- skill-enhancers
+- analyzer
+disallowedTools: []
+skills: []
+background: false
+# ── upgrade levers — uncomment + set when tuning this agent ──
+# effort: high            # reasoning depth: low/medium/high/xhigh/max (omit = inherit session)
+# maxTurns: 50            # cap the agentic loop (omit = engine default)
+# memory: project         # persistent scope: user/project/local (omit = ephemeral)
+# isolation: worktree     # run in an isolated git worktree
+# initialPrompt: "…"      # seed the agent's first turn
+# hooks / mcpServers / permissionMode → set at the PLUGIN level, not on a plugin agent
 ---
-
 # Post-hoc Analyzer Agent
 
 Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
@@ -54,6 +81,7 @@ You receive these parameters in your prompt:
 ### Step 4: Analyze Instruction Following
 
 For each transcript, evaluate:
+
 - Did the agent follow the skill's explicit instructions?
 - Did the agent use the skill's provided tools/scripts?
 - Were there missed opportunities to leverage skill content?
@@ -64,6 +92,7 @@ Score instruction following 1-10 and note specific issues.
 ### Step 5: Identify Winner Strengths
 
 Determine what made the winner better:
+
 - Clearer instructions that led to better behavior?
 - Better scripts/tools that produced better output?
 - More comprehensive examples that guided edge cases?
@@ -74,6 +103,7 @@ Be specific. Quote from skills/transcripts where relevant.
 ### Step 6: Identify Loser Weaknesses
 
 Determine what held the loser back:
+
 - Ambiguous instructions that led to suboptimal choices?
 - Missing tools/scripts that forced workarounds?
 - Gaps in edge case coverage?
@@ -82,6 +112,7 @@ Determine what held the loser back:
 ### Step 7: Generate Improvement Suggestions
 
 Based on the analysis, produce actionable suggestions for improving the loser skill:
+
 - Specific instruction changes to make
 - Tools/scripts to add or modify
 - Examples to include
@@ -216,6 +247,7 @@ You receive these parameters in your prompt:
 ### Step 2: Analyze Per-Assertion Patterns
 
 For each expectation across all runs:
+
 - Does it **always pass** in both configurations? (may not differentiate skill value)
 - Does it **always fail** in both configurations? (may be broken or beyond capability)
 - Does it **always pass with skill but fail without**? (skill clearly adds value here)
@@ -225,6 +257,7 @@ For each expectation across all runs:
 ### Step 3: Analyze Cross-Eval Patterns
 
 Look for patterns across evals:
+
 - Are certain eval types consistently harder/easier?
 - Do some evals show high variance while others are stable?
 - Are there surprising results that contradict expectations?
@@ -232,6 +265,7 @@ Look for patterns across evals:
 ### Step 4: Analyze Metrics Patterns
 
 Look at time_seconds, tokens, tool_calls:
+
 - Does the skill significantly increase execution time?
 - Is there high variance in resource usage?
 - Are there outlier runs that skew the aggregates?
@@ -239,11 +273,13 @@ Look at time_seconds, tokens, tool_calls:
 ### Step 5: Generate Notes
 
 Write freeform observations as a list of strings. Each note should:
+
 - State a specific observation
 - Be grounded in the data (not speculation)
 - Help the user understand something the aggregate metrics don't show
 
 Examples:
+
 - "Assertion 'Output is a PDF file' passes 100% in both configurations - may not differentiate skill value"
 - "Eval 3 shows high variance (50% ± 40%) - run 2 had an unusual failure that may be flaky"
 - "Without-skill runs consistently fail on table extraction expectations (0% pass rate)"
@@ -267,12 +303,14 @@ Save notes to `{output_path}` as a JSON array of strings:
 ## Guidelines
 
 **DO:**
+
 - Report what you observe in the data
 - Be specific about which evals, expectations, or runs you're referring to
 - Note patterns that aggregate metrics would hide
 - Provide context that helps interpret the numbers
 
 **DO NOT:**
+
 - Suggest improvements to the skill (that's for the improvement step, not benchmarking)
 - Make subjective quality judgments ("the output was good/bad")
 - Speculate about causes without evidence

package/skills/skill-creator/agents/comparator.md

@@ -1,8 +1,35 @@
 ---
 name: comparator
 description: Compare two outputs blindly without knowing which skill produced them
+tools:
+- Read
+- Write
+- Edit
+- Bash
+- Glob
+- Grep
+- WebFetch
+- WebSearch
+- Task
+- TodoWrite
+model: sonnet
+color: red
+version: 1.0.0
+author: Jeremy Longshore <jeremy@intentsolutions.io>
+tags:
+- skill-enhancers
+- comparator
+disallowedTools: []
+skills: []
+background: false
+# ── upgrade levers — uncomment + set when tuning this agent ──
+# effort: high            # reasoning depth: low/medium/high/xhigh/max (omit = inherit session)
+# maxTurns: 50            # cap the agentic loop (omit = engine default)
+# memory: project         # persistent scope: user/project/local (omit = ephemeral)
+# isolation: worktree     # run in an isolated git worktree
+# initialPrompt: "…"      # seed the agent's first turn
+# hooks / mcpServers / permissionMode → set at the PLUGIN level, not on a plugin agent
 ---
-
 # Blind Comparator Agent
 
 Compare two outputs WITHOUT knowing which skill produced them.
@@ -44,6 +71,7 @@ You receive these parameters in your prompt:
 Based on the task, generate a rubric with two dimensions:
 
 **Content Rubric** (what the output contains):
+
 | Criterion | 1 (Poor) | 3 (Acceptable) | 5 (Excellent) |
 |-----------|----------|----------------|---------------|
 | Correctness | Major errors | Minor errors | Fully correct |
@@ -51,6 +79,7 @@ Based on the task, generate a rubric with two dimensions:
 | Accuracy | Significant inaccuracies | Minor inaccuracies | Accurate throughout |
 
 **Structure Rubric** (how the output is organized):
+
 | Criterion | 1 (Poor) | 3 (Acceptable) | 5 (Excellent) |
 |-----------|----------|----------------|---------------|
 | Organization | Disorganized | Reasonably organized | Clear, logical structure |
@@ -58,6 +87,7 @@ Based on the task, generate a rubric with two dimensions:
 | Usability | Difficult to use | Usable with effort | Easy to use |
 
 Adapt criteria to the specific task. For example:
+
 - PDF form → "Field alignment", "Text readability", "Data placement"
 - Document → "Section structure", "Heading hierarchy", "Paragraph flow"
 - Data output → "Schema correctness", "Data types", "Completeness"

package/skills/skill-creator/agents/grader.md

@@ -1,8 +1,35 @@
 ---
 name: grader
 description: Evaluate expectations against execution transcripts and outputs
+tools:
+- Read
+- Write
+- Edit
+- Bash
+- Glob
+- Grep
+- WebFetch
+- WebSearch
+- Task
+- TodoWrite
+model: sonnet
+color: pink
+version: 1.0.0
+author: Jeremy Longshore <jeremy@intentsolutions.io>
+tags:
+- skill-enhancers
+- grader
+disallowedTools: []
+skills: []
+background: false
+# ── upgrade levers — uncomment + set when tuning this agent ──
+# effort: high            # reasoning depth: low/medium/high/xhigh/max (omit = inherit session)
+# maxTurns: 50            # cap the agentic loop (omit = engine default)
+# memory: project         # persistent scope: user/project/local (omit = ephemeral)
+# isolation: worktree     # run in an isolated git worktree
+# initialPrompt: "…"      # seed the agent's first turn
+# hooks / mcpServers / permissionMode → set at the PLUGIN level, not on a plugin agent
 ---
-
 # Grader Agent
 
 Evaluate expectations against an execution transcript and outputs.
@@ -66,6 +93,7 @@ This catches issues that predefined expectations might miss.
 ### Step 5: Read User Notes
 
 If `{outputs_dir}/user_notes.md` exists:
+
 1. Read it and note any uncertainties or issues flagged by the executor
 2. Include relevant concerns in the grading output
 3. These may reveal problems even when expectations pass
@@ -77,6 +105,7 @@ After grading, consider whether the evals themselves could be improved. Only sur
 Good suggestions test meaningful outcomes — assertions that are hard to satisfy without actually doing the work correctly. Think about what makes an assertion *discriminating*: it passes when the skill genuinely succeeds and fails when it doesn't.
 
 Suggestions worth raising:
+
 - An assertion that passed but would also pass for a clearly wrong output (e.g., checking filename existence but not file content)
 - An important outcome you observed — good or bad — that no assertion covers at all
 - An assertion that can't actually be verified from the available outputs
@@ -90,11 +119,13 @@ Save results to `{outputs_dir}/../grading.json` (sibling to outputs_dir).
 ## Grading Criteria
 
 **PASS when**:
+
 - The transcript or outputs clearly demonstrate the expectation is true
 - Specific evidence can be cited
 - The evidence reflects genuine substance, not just surface compliance (e.g., a file exists AND contains correct content, not just the right filename)
 
 **FAIL when**:
+
 - No evidence found for the expectation
 - Evidence contradicts the expectation
 - The expectation cannot be verified from available information

package/skills/skill-creator/eval-viewer/generate_review.py

@@ -32,9 +32,32 @@ METADATA_FILES = {"transcript.md", "user_notes.md", "metrics.json"}
 
 # Extensions we render as inline text
 TEXT_EXTENSIONS = {
-    ".txt", ".md", ".json", ".csv", ".py", ".js", ".ts", ".tsx", ".jsx",
-    ".yaml", ".yml", ".xml", ".html", ".css", ".sh", ".rb", ".go", ".rs",
-    ".java", ".c", ".cpp", ".h", ".hpp", ".sql", ".r", ".toml",
+    ".txt",
+    ".md",
+    ".json",
+    ".csv",
+    ".py",
+    ".js",
+    ".ts",
+    ".tsx",
+    ".jsx",
+    ".yaml",
+    ".yml",
+    ".xml",
+    ".html",
+    ".css",
+    ".sh",
+    ".rb",
+    ".go",
+    ".rs",
+    ".java",
+    ".c",
+    ".cpp",
+    ".h",
+    ".hpp",
+    ".sql",
+    ".r",
+    ".toml",
 }
 
 # Extensions we render as inline images
@@ -224,9 +247,7 @@ def load_previous_iteration(workspace: Path) -> dict[str, dict]:
         try:
             data = json.loads(feedback_path.read_text())
             feedback_map = {
-                r["run_id"]: r["feedback"]
-                for r in data.get("reviews", [])
-                if r.get("feedback", "").strip()
+                r["run_id"]: r["feedback"] for r in data.get("reviews", []) if r.get("feedback", "").strip()
             }
         except (json.JSONDecodeError, OSError, KeyError):
             pass
@@ -285,12 +306,15 @@ def generate_html(
 # HTTP server (stdlib only, zero dependencies)
 # ---------------------------------------------------------------------------
 
+
 def _kill_port(port: int) -> None:
     """Kill any process listening on the given port."""
     try:
         result = subprocess.run(
             ["lsof", "-ti", f":{port}"],
-            capture_output=True, text=True, timeout=5,
+            capture_output=True,
+            text=True,
+            timeout=5,
         )
         for pid_str in result.stdout.strip().split("\n"):
             if pid_str.strip():
@@ -305,6 +329,7 @@ def _kill_port(port: int) -> None:
     except FileNotFoundError:
         print("Note: lsof not found, cannot check if port is in use", file=sys.stderr)
 
+
 class ReviewHandler(BaseHTTPRequestHandler):
     """Serves the review HTML and handles feedback saves.
 
@@ -390,15 +415,22 @@ def main() -> None:
     parser.add_argument("--port", "-p", type=int, default=3117, help="Server port (default: 3117)")
     parser.add_argument("--skill-name", "-n", type=str, default=None, help="Skill name for header")
     parser.add_argument(
-        "--previous-workspace", type=Path, default=None,
+        "--previous-workspace",
+        type=Path,
+        default=None,
         help="Path to previous iteration's workspace (shows old outputs and feedback as context)",
     )
     parser.add_argument(
-        "--benchmark", type=Path, default=None,
+        "--benchmark",
+        type=Path,
+        default=None,
         help="Path to benchmark.json to show in the Benchmark tab",
     )
     parser.add_argument(
-        "--static", "-s", type=Path, default=None,
+        "--static",
+        "-s",
+        type=Path,
+        default=None,
         help="Write standalone HTML to this path instead of starting a server",
     )
     args = parser.parse_args()
@@ -447,8 +479,8 @@ def main() -> None:
         port = server.server_address[1]
 
     url = f"http://localhost:{port}"
-    print(f"\n  Eval Viewer")
-    print(f"  ─────────────────────────────────")
+    print("\n  Eval Viewer")
+    print("  ─────────────────────────────────")
     print(f"  URL:       {url}")
     print(f"  Workspace: {workspace}")
     print(f"  Feedback:  {feedback_path}")
@@ -456,7 +488,7 @@ def main() -> None:
         print(f"  Previous:  {args.previous_workspace} ({len(previous)} runs)")
     if benchmark_path:
         print(f"  Benchmark: {benchmark_path}")
-    print(f"\n  Press Ctrl+C to stop.\n")
+    print("\n  Press Ctrl+C to stop.\n")
 
     webbrowser.open(url)