CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

soul-guardian

Drift detection + baseline integrity guard for agent workspace files with automatic alerting support

Quality

62%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./skills/soul-guardian/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

57%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A well-structured, highly executable skill body whose commands are nearly all copy-paste ready and accurate, with honest security-model framing. Its weaknesses are the inlined standalone-install verification script, a broken demo.sh reference, orphaned bundle scripts with no navigation to them, and the absence of a validation checkpoint before enabling destructive auto-restore.

Suggestions

Move the ~75-line Release Artifact Verification script into a reference file (e.g. references/verify-release.md) and keep a one-line pointer plus the fingerprint in SKILL.md.

Fix the bundle mismatch: either add scripts/demo.sh or remove the Demo section, and add short pointers to the existing install_launchd_plist.py and onboard_state_dir.py so the launchctl/openclaw mentions resolve to real files.

Add an explicit validation checkpoint to the Quick Start before enabling monitoring — e.g. run `check` and confirm baselines are correct, and warn that restore mode overwrites drifted files — since the destructive-operation rubric caps workflow clarity at 3 without it.

DimensionReasoningScore

Conciseness

Most sections (Quick Start, Commands, policy table, Security Model, Troubleshooting) are tight and instruction-first, but the ~75-line release-verification bash block (lines 37-108) is inline bulk relevant only to standalone installs, and the "That's it! Soul Guardian will now: ✅ ..." checklist restates the policy table. Fits anchor 3 (mostly efficient, could be tightened); not 2 since there is little concept-explanation padding.

3 / 5

Actionability

Nearly all commands are copy-paste ready with full flags (init, check --output-format alert, watch --interval 30, approve --file ... --actor ... --note ..., status, verify-audit), and all subcommands were verified to exist in scripts/soul_guardian.py. It falls short of 5 because `bash skills/soul-guardian/scripts/demo.sh` (line 238) references a script that does not exist in the bundle, and the enable-monitoring step only says "This shows you how to add monitoring" without stating the outcome.

4 / 5

Workflow Clarity

The Quick Start is a clear 3-step sequence (init -> enable-monitoring -> heartbeat check), but the skill's core operation is destructive — restore mode intentionally overwrites drifted files — and the workflow has no explicit pre-enable validation checkpoint (e.g. verifying baselines or a dry-run check) before destructive behavior is armed. Per the rubric's cap for destructive operations without validation, workflow_clarity cannot exceed 3; the verify-audit/status commands exist but are never sequenced into the workflow as checkpoints.

3 / 5

Progressive Disclosure

The body has good section structure (Quick Start, Commands, Alert Format, Security Model, Troubleshooting), but scored against the actual bundle: the 75-line release-verification script is inlined where a one-level-deep reference file belongs; `demo.sh` is referenced but missing from scripts/; and the bundled `install_launchd_plist.py` and `onboard_state_dir.py` (which the Operational Notes' launchctl/openclaw mentions allude to) are never pointed to, so navigation to a third of the bundle is undiscoverable. Fits anchor 3 (some structure, content that should be separate is inline, references not clearly signaled); the intact header organization keeps it above anchor 2.

3 / 5

Total

13

/

20

Passed

Description

53%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A terse, third-person description that states concrete capabilities for a genuinely distinct niche, but it reads as a feature label rather than a trigger-rich description. The missing "Use when..." clause and the absence of the natural file names users would mention (SOUL.md, AGENTS.md) keep it at the midpoint across most dimensions.

Suggestions

Add an explicit trigger clause, e.g. "Use when protecting agent core files (SOUL.md, AGENTS.md), detecting unauthorized modifications, or setting up file-integrity baselines and alerting."

Replace label-style phrasing with concrete operation verbs so capabilities are enumerated: "Detects drift, restores files to approved baselines, and alerts on unauthorized changes to agent core files."

Include the natural trigger terms users would actually say — SOUL.md, AGENTS.md, unauthorized changes, file integrity — instead of jargon-only terms like "baseline integrity guard".

DimensionReasoningScore

Specificity

Names the domain ("agent workspace files") and 2-3 actions ("drift detection", "baseline integrity guard", "automatic alerting"), but phrasing like "+ ... guard ... with ... support" is compact to the point of being label-like rather than enumerating concrete operations. This matches anchor 3 (domain plus 1-2 concrete actions, not comprehensive); it is above anchor 2 because real, specific capabilities are named, and below anchor 4 because no operational verbs like restore, verify, or approve appear.

3 / 5

Completeness

The description clearly answers "what" (drift detection + baseline integrity guard + alerting for agent workspace files) but contains no "Use when..." clause or equivalent trigger guidance. Per the judging guideline, a missing explicit trigger caps completeness at 3; it is not 4 because the "when" is entirely absent rather than merely imprecise.

3 / 5

Trigger Term Quality

Terms like "drift detection", "baseline integrity", and "alerting" are domain-relevant but jargon-leaning, and the natural phrases a user would actually say ("SOUL.md", "AGENTS.md", "unauthorized changes", "file integrity", "restore my agent's core files") are absent. This fits anchor 3 (some relevant keywords, missing common variations/synonyms); not anchor 4, which requires good coverage of natural user terms.

3 / 5

Distinctiveness Conflict Risk

"Drift detection + baseline integrity guard for agent workspace files" carves out a fairly distinct niche (agent core-file protection) with minor overlap risk against generic file-integrity, backup, or security-monitoring skills. Fits anchor 4 (mostly distinct); not 5 because it omits its most distinctive triggers (SOUL.md, AGENTS.md), which would make conflict risk minimal.

4 / 5

Total

13

/

20

Passed

Validation

81%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 13 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

metadata_version

'metadata.version' is missing

Warning

metadata_field

'metadata' should map string keys to string values

Warning

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

Total

13

/

16

Passed

Repository

prompt-security/clawsec

Path

skills/soul-guardian/SKILL.md

Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.