SKILLEMALL.ai
GitHub ↗ RU Sign in

DD loki-mode

Runs an autonomous software factory that knows what it is supposed to deliver and proves it did, with a signed receipt. Use when the user says Loki Mode or asks to build, fix or verify software autonomously.

The skillemall take

Promises an autonomous software factory with signed delivery receipts. In reality: 80 files and 49 scripts with half the links broken. Audits found 14 medium and low-severity issues—logic mismatches, undocumented dependencies, potential context leaks. No model testing, no sandbox runs. Quality score 57, process score 45—raw prototype material.

The catch: "autonomous" here means wandering without a map. No guarantee the agent understands what it should deliver, and a signed receipt sounds like wishful thinking. Install only if you're prepared to debug live and hand-verify every run.

asklokesh/claudeskill-loki-mode Agent Skills author: asklokesh 80 files · 49 scripts body ≈ 7 424 tokens Open the sourcegithub.com↗ analyzed 2 d ago

Runs an autonomous software factory that knows what it is supposed to deliver and proves it did, with a signed receipt.

As a process D 45/100 · Unfinished process — References files that are not bundled: skills/model-selection.md, skills/agents.md, skills/testing.md

AnalyzerGitHubDockerAI and agentsInfrastructuretype and topics are labelled automatically from the skill text
Runs in: Claude / Claude Code ChatGPT / Codex Gemini CLI Cursor GitHub Copilot / VS Code Qwen Code Kimi Code CLI Z.ai (GLM) DeepSeek (Deep Code) Mistral Vibe LangChain (deepagents) OpenClaw Goose OpenCode
JSON
Technical rating
D
65/100
safety, quality, tests
Safety 60%
70
Quality 40%
57
Run on models
none yet
Process rating
D
45/100
Unfinished process
References files that are not bundled: skills/model-selection.md, skills/agents.md, skills/testing.md
Tools and files w 18
0
Result and completion w 14
0
Progress reporting w 2
0
the three weakest of ten parameters · all ten

What is at stake

Medium-severity findings: the skill is probably honest, but read what alarmed the scanner.

Dangerous commands medium severity

Below is the worst case for this category. The finding here is medium: the guard saw a sign, not a proof.

If you install

The skill contains commands that delete files, rewrite disks or execute code fetched from the network. The agent may run them without asking if it believes the instructions require it.

For the author

Replace destructive commands with safe equivalents that ask for confirmation, scope them to one folder, and stop piping curl into a shell: pin a version and a checksum.

Exfiltration medium severity

Below is the worst case for this category. The finding here is medium: the guard saw a sign, not a proof.

If you install

The instructions or scripts send data out: environment variables, keys, file contents, chat history. You may never notice, because the agent performs the upload as "part of the task".

For the author

If the upload is not needed for the task, remove it: catalogs flag such skills and delist them. If it is needed, name the destination explicitly, say what leaves the machine, and give the user a switch.

How to improve

  1. The SKILL.md body is over 5,000 tokens: move reference detail into references/ and load it when needed.
  2. The text references files that are not there: add them or drop the references.
For the model run — optional
  • Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
  • A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.

Guard findings · 14

✓ No critical or high findings

Medium and low: 14
  • medium Dangerous commands cmd-pipe-to-shell-known-host api/README.md:25
    Pipe-to-shell installer from a well-known host (still executes remote code)
    curl -fsSL https://deno.land/install.sh | sh
  • medium Exfiltration net-credential-use api/README.md:65
    Credential used in a network call (verify the destination is the intended service)
    curl -H "Authorization: Bearer $LOKI_API_TOKEN" http://serv…374/health
  • medium Dangerous commands cmd-destructive-fs autonomy/config.example.yaml:54
    Destructive filesystem command (wipes root/home/drive) (detector / deny-list definition; string literal in code, not executed; test fixture / example file)
    blocked_commands: "rm -rf /,dd if=,mkfs,:(){ :|:& };:"
    detectorcode literalfixture
  • medium Exfiltration exfil-read-secret-files autonomy/docker-run.sh:77
    Reads credential / secret files (quoted — discussed, not commanded)
    raw="$(cat "${HOME}/.claude/.credentials.json" 2>/dev/null || true)"
    quoted
  • low Dangerous commands cmd-pipe-to-shell-known-host .github/CONTRIBUTING.md:9
    Pipe-to-shell installer from a well-known host (still executes remote code) (quoted — discussed, not commanded)
    - **Bun 1.3+** (`brew install oven-sh/bun/bun` or `curl -fsSL https://bun.sh/install | bash`) -- powers the TypeScript runner under `loki-ts/`
    quoted
  • low Secrets in code secret-high-entropy-token assets/welcome/welcome.html:262
    High-entropy token-like string (may be an id, hash or a credential) (quoted — discussed, not commanded)
    var POSTHOG_KEY = "phc_…s87";
    quoted
  • low Dangerous commands cmd-background-process autonomy/app-runner.sh:277
    Starts a background / autostarted process (code comment)
    # from the root. The root PID itself is NOT included. Used by the non-setsid stop
    comment
  • low Dangerous commands cmd-background-process autonomy/app-runner.sh:278
    Starts a background / autostarted process (negated — the text forbids it)
    # fallback (BUG 1): the app is started as `( ... ) &` WITHOUT setsid, so on stock
    negated
  • low Dangerous commands cmd-background-process autonomy/app-runner.sh:346
    Starts a background / autostarted process (code comment)
    # a process group) because in the non-setsid path the app inherits the
    comment
  • low Dangerous commands cmd-background-process autonomy/app-runner.sh:364
    Starts a background / autostarted process (code comment)
    # Used by the non-setsid stop grace-wait so a deep worker that outlived the main
    comment
  • low Dangerous commands cmd-background-process autonomy/app-runner.sh:1282
    Starts a background / autostarted process
    if command -v setsid >/dev/null 2>&1; then
  • low Dangerous commands cmd-background-process autonomy/completion-council.sh:4886
    Starts a background / autostarted process
    disown 2>/dev/null || true
  • low Dangerous commands cmd-background-process autonomy/docker-run.sh:215
    Starts a background / autostarted process (negated — the text forbids it)
    # The container IS the session boundary, so the runner must NOT setsid into
    negated
  • low Dangerous commands cmd-background-process autonomy/docker-run.sh:216
    Starts a background / autostarted process (code comment)
    # a new, detached session: setsid detach inside a `--rm` container makes the
    comment

Files scanned: 79. Evidence is masked. Grey chips explain why severity was lowered.

Against the Agent Skills spec

  • warning body-long SKILL.md body ≈ 7424 tokens (recommended < 5000); move details to references/
  • warning missing-ref reference to a missing file: skills/model-selection.md
  • warning missing-ref reference to a missing file: skills/agents.md
  • warning missing-ref reference to a missing file: skills/testing.md
  • warning missing-ref reference to a missing file: skills/quality-gates.md
  • warning missing-ref reference to a missing file: skills/production.md
  • warning missing-ref reference to a missing file: skills/troubleshooting.md
  • warning missing-ref reference to a missing file: skills/patterns-advanced.md
  • warning missing-ref reference to a missing file: skills/artifacts.md
  • warning missing-ref reference to a missing file: skills/parallel-workflows.md
  • warning missing-ref reference to a missing file: skills/github-integration.md
  • warning missing-ref reference to a missing file: skills/providers.md
  • warning missing-ref reference to a missing file: skills/openspec-integration.md
  • warning missing-ref reference to a missing file: skills/mirofish-integration.md
  • warning missing-ref reference to a missing file: skills/documentation.md
  • warning missing-ref reference to a missing file: skills/magic-modules.md
  • warning missing-ref reference to a missing file: skills/healing.md
  • warning missing-ref reference to a missing file: skills/compound-learning.md
  • warning missing-ref reference to a missing file: skills/memory.md
  • warning missing-ref reference to a missing file: skills/factory-operations.md
  • warning missing-ref reference to a missing file: skills/release-cadence.md
  • warning missing-ref reference to a missing file: skills/extending.md
  • warning missing-ref reference to a missing file: references/sdk-mode.md
  • warning missing-ref reference to a missing file: references/cursor-learnings.md
  • warning missing-ref reference to a missing file: references/memory-system.md
  • warning missing-ref reference to a missing file: references/core-workflow.md
  • note edit-residue the text marks something as outdated (lines 4, 14, 16, 204, 233, 257): check that old rules are not kept next to new ones — the full check reads the text for contradictions

Process rating: all ten parameters 45/100

Will not run. References files that are not bundled: skills/model-selection.md, skills/agents.md, skills/testing.md
  • 0Tools and files. 25 referenced file(s) missing: skills/model-selection.md, skills/agents.md, skills/testing.md
  • 0Result and completion. Does not say what the result is
  • 0Progress reporting. Says nothing while it works
  • 30Running it twice. 15 mutating operations with no state check
  • 40Consistency. Frontmatter name (loki-mode) differs from the folder (claudeskill-loki-mode)
  • 50Failures and branches. 0 branches, has a failure section
  • 70When it triggers. States when to use, but not when not to
  • 70Inputs and preconditions. Inputs and preconditions are listed
  • 70Execution cost. Instruction body is 7424 tokens
  • 100Steps. 27 steps
  • medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
  • low 20 top-level sections: this looks like several domains in one skill
  • low The response is described with custom markup (10 tags): a typed call is more reliable

Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.

Quality signals

  • +5Description has no quoted example phrases that should trigger the skill
  • +4Description does not say when NOT to use the skill (false activations)
  • +3Output format is not stated: the model decides each time
  • -5TODO / placeholder text left in the skill
  • +1No license
  • +2Single-language instructions
  • +3Description length 207: enough signal without eating the budget
  • +4Structure: 23 headings
  • +3Step-by-step instructions: 27 items
  • +4Has examples (8 code blocks)

Quality base 70; lint remarks subtract, signals add up to 100. Result: 57.