f153f969a0
Add /ultraresearch-local for structured research combining local codebase analysis with external knowledge via parallel agent swarms. Produces research briefs with triangulation, confidence ratings, and source quality assessment. New command: /ultraresearch-local with modes --quick, --local, --external, --fg. New agents: research-orchestrator (opus), docs-researcher, community-researcher, security-researcher, contrarian-researcher, gemini-bridge (all sonnet). New template: research-brief-template.md. Integration: --research flag in /ultraplan-local accepts pre-built research briefs (up to 3), enriches the interview and exploration phases. Planning orchestrator cross-references brief findings during synthesis. Design principle: Context Engineering — right information to right agent at right time. Research briefs are structured artifacts in the pipeline: ultraresearch → brief → ultraplan --research → plan → ultraexecute. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
189 lines
6.6 KiB
Markdown
189 lines
6.6 KiB
Markdown
# Security Posture Scorecard
|
||
|
||
<!--
|
||
TEMPLATE USAGE
|
||
This is a reference document describing the expected output structure for `/security posture`.
|
||
Agents use this as a formatting guide for a quick, human-readable posture assessment.
|
||
Fill every section with real observations. Do NOT output placeholder text.
|
||
This is a lightweight assessment — not a full audit. Aim for signal over exhaustiveness.
|
||
-->
|
||
|
||
---
|
||
|
||
## Header
|
||
|
||
**Project:** [Name of the project or directory assessed]
|
||
**Assessment date:** [ISO 8601 — e.g. 2026-02-19]
|
||
**Assessed by:** llm-security plugin v[X.X] — posture-assessor-agent
|
||
**Mode:** Quick assessment (for full audit run `/security audit`)
|
||
|
||
---
|
||
|
||
## Overall Score
|
||
|
||
**[N] / 9 categories covered**
|
||
|
||
```
|
||
[==========> ] [N]/9 [Rating label]
|
||
```
|
||
|
||
Rating labels by score:
|
||
- 9/9 — Fully secured
|
||
- 7–8/9 — Well secured
|
||
- 5–6/9 — Partially secured
|
||
- 3–4/9 — Significant gaps
|
||
- 0–2/9 — Critical gaps
|
||
|
||
**One-line verdict:** [e.g. "3 gaps require immediate attention before this plugin is safe for production use."]
|
||
|
||
---
|
||
|
||
## Category Scorecard
|
||
|
||
Each category is marked with one of four indicators:
|
||
- COVERED — Control is in place and effective
|
||
- PARTIAL — Control exists but has gaps
|
||
- GAP — Control is absent or broken
|
||
- N/A — Not applicable to this project
|
||
|
||
| # | Category | Status | Notes |
|
||
|---|----------|--------|-------|
|
||
| 1 | Deny-First Configuration | [COVERED / PARTIAL / GAP / N/A] | [1–2 lines: what is in place or what is missing] |
|
||
| 2 | Secrets Protection | [COVERED / PARTIAL / GAP / N/A] | [1–2 lines] |
|
||
| 3 | Path Guarding | [COVERED / PARTIAL / GAP / N/A] | [1–2 lines] |
|
||
| 4 | MCP Server Trust | [COVERED / PARTIAL / GAP / N/A] | [1–2 lines] |
|
||
| 5 | Destructive Command Blocking | [COVERED / PARTIAL / GAP / N/A] | [1–2 lines] |
|
||
| 6 | Sandbox Configuration | [COVERED / PARTIAL / GAP / N/A] | [1–2 lines] |
|
||
| 7 | Human Review Requirements | [COVERED / PARTIAL / GAP / N/A] | [1–2 lines] |
|
||
| 8 | Skill and Plugin Sources | [COVERED / PARTIAL / GAP / N/A] | [1–2 lines] |
|
||
| 9 | Session Isolation | [COVERED / PARTIAL / GAP / N/A] | [1–2 lines] |
|
||
|
||
---
|
||
|
||
## Category Detail
|
||
|
||
### 1. Deny-First Configuration
|
||
[What deny-first controls were found, or what is missing. Reference specific config files if present.]
|
||
|
||
### 2. Secrets Protection
|
||
[Describe hook coverage, `.gitignore` patterns, and any hardcoded secrets found. Redact actual values.]
|
||
|
||
### 3. Path Guarding
|
||
[Which sensitive paths are guarded. List any unprotected paths that should be blocked.]
|
||
|
||
### 4. MCP Server Trust
|
||
[Number of MCP servers found. Trust status for each: verified / unverified / local-only.]
|
||
|
||
### 5. Destructive Command Blocking
|
||
[Hook presence. Which destructive patterns are blocked. Any patterns that are missing.]
|
||
|
||
### 6. Sandbox Configuration
|
||
[Network access scope, file system scope, any overly permissive settings found.]
|
||
|
||
### 7. Human Review Requirements
|
||
[Whether high-impact operations require confirmation. Examples of confirmation gates found or absent.]
|
||
|
||
### 8. Skill and Plugin Sources
|
||
[Number of plugins/skills. Source verification status. Any plugins from unverified sources.]
|
||
|
||
### 9. Session Isolation
|
||
[How context is shared between agents and sessions. Any cross-session state leakage risks.]
|
||
|
||
---
|
||
|
||
## Top 3 Recommendations
|
||
|
||
These are the highest-impact actions to improve posture, ordered by urgency.
|
||
|
||
**1. [Title of recommendation]**
|
||
Category: [Category name]
|
||
Risk: [What could happen if not addressed]
|
||
Action: [Specific step to take]
|
||
Effort: [Low / Medium / High]
|
||
|
||
**2. [Title of recommendation]**
|
||
Category: [Category name]
|
||
Risk: [What could happen if not addressed]
|
||
Action: [Specific step to take]
|
||
Effort: [Low / Medium / High]
|
||
|
||
**3. [Title of recommendation]**
|
||
Category: [Category name]
|
||
Risk: [What could happen if not addressed]
|
||
Action: [Specific step to take]
|
||
Effort: [Low / Medium / High]
|
||
|
||
---
|
||
|
||
## Quick Wins
|
||
|
||
Things that can be fixed in under 5 minutes with no architectural changes.
|
||
|
||
- [ ] [Quick win action — e.g. "Add `.env` to `.gitignore`"]
|
||
- [ ] [Quick win action — e.g. "Enable `pre-edit-secrets` hook from claude-code-essentials"]
|
||
- [ ] [Quick win action — e.g. "Remove hardcoded API key on line 42 of config.json"]
|
||
|
||
> If no quick wins are identified, write: "No quick wins identified — improvements require architectural changes."
|
||
|
||
---
|
||
|
||
## Baseline Comparison
|
||
|
||
What a fully secured Claude Code project looks like vs. this project.
|
||
|
||
| Category | Fully Secured | This Project |
|
||
|----------|--------------|--------------|
|
||
| Deny-First Configuration | `defaultPermissionLevel: deny` in settings | [Current state] |
|
||
| Secrets Protection | Hook active + `.env` gitignored + no hardcoded secrets | [Current state] |
|
||
| Path Guarding | `pre-write-pathguard` hook blocks sensitive paths | [Current state] |
|
||
| MCP Server Trust | All servers verified, minimal scope, auth required | [Current state] |
|
||
| Destructive Command Blocking | `pre-bash-destructive` hook with comprehensive patterns | [Current state] |
|
||
| Sandbox Configuration | Network and filesystem access scoped to project | [Current state] |
|
||
| Human Review Requirements | Confirmation gates before irreversible operations | [Current state] |
|
||
| Skill and Plugin Sources | All plugins from verified sources, minimal permissions | [Current state] |
|
||
| Session Isolation | No cross-session state leakage, minimal context sharing | [Current state] |
|
||
|
||
**Gap summary:** [N] of 9 categories match the fully secured baseline. [N] have partial coverage. [N] have no coverage.
|
||
|
||
---
|
||
|
||
## Footer
|
||
|
||
| Field | Value |
|
||
|-------|-------|
|
||
| llm-security version | [e.g. 0.1.0] |
|
||
| Assessment engine | posture-assessor-agent |
|
||
| Full audit command | `/security audit` |
|
||
| Report generated | [ISO 8601 timestamp] |
|
||
|
||
---
|
||
|
||
<!--
|
||
SCORING LOGIC (for agents filling in this template)
|
||
|
||
Score = number of categories with status COVERED (not PARTIAL, GAP, or N/A).
|
||
N/A categories are excluded from the denominator AND the score.
|
||
|
||
Score display denominator = 9 - (count of N/A categories)
|
||
|
||
Progress bar fill = round((score / denominator) * 10) blocks out of 10
|
||
|
||
Rating labels:
|
||
100% → Fully secured
|
||
78–99% → Well secured
|
||
56–77% → Partially secured
|
||
34–55% → Significant gaps
|
||
0–33% → Critical gaps
|
||
|
||
TOP 3 SELECTION LOGIC
|
||
Select the 3 GAP or PARTIAL categories with the highest potential impact:
|
||
Priority 1: GAP in Secrets Protection, Deny-First, or Destructive Blocking
|
||
Priority 2: GAP in MCP Trust, Path Guarding, or Sandbox
|
||
Priority 3: PARTIAL in any category, or GAP in Human Review / Session Isolation
|
||
|
||
QUICK WINS CRITERIA
|
||
A quick win qualifies if:
|
||
- It can be resolved with a single file edit or config change
|
||
- It requires no new dependencies or architectural decisions
|
||
- Estimated time: under 5 minutes
|
||
-->
|