BF research-mode
Run durable long-running research in OpenClaw using isolated cron iterations, persistent state, bounded execution, and milestone updates. Use when the user wants background research that can continue for hours or days, pause/resume/stop cleanly, accumulate sources/findings over time, and produce a final report instead of a single one-shot answer.
Run durable long-running research in OpenClaw using isolated cron iterations, persistent state, bounded execution, and milestone updates.
As a process F 52/100 · Will not run — References files that are not bundled: scripts/check_research_mode.sh, scripts/selftest/
What is at stake
Medium-severity findings: the skill is probably honest, but read what alarmed the scanner.
Below is the worst case for this category. The finding here is medium: the guard saw a sign, not a proof.
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.
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.
How to improve
- The SKILL.md body is over 5,000 tokens: move reference detail into references/ and load it when needed.
- The text references files that are not there: add them or drop the references.
- 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 · 4
✓ No critical or high findings
Medium and low: 4
-
medium Dangerous commands
cmd-pipe-to-shellRELEASING.md:100Downloads and executes remote code from an unrecognised host (pipe to shell) (quoted — discussed, not commanded)- [ ] Audit dependency instructions for `curl|sh`, unpinned remote scripts, or blind package installs.
quoted -
medium Dangerous commands
cmd-pipe-to-shellRELEASING.md:238Downloads and executes remote code from an unrecognised host (pipe to shell) (quoted — discussed, not commanded)- [ ] Проверить инструкции по зависимостям на `curl|sh`, удалённые скрипты без закреплённой версии и слепую установку пакетов.
quoted -
low Risky intent
intent-offensive-securityexamples/rag-eval-tooling-matrix/research-trace/iterations/004.md:29Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (test fixture / example file)- [Giskard AI Red Teaming and LLM Security Platform](https://www.giskard.ai/) — Official Giskard Hub enterprise controls evidence: data residency, RBAC, audit trails, identity-provider integration, co
fixture -
low Risky intent
intent-offensive-securityexamples/rag-eval-tooling-matrix/sources.md:140Offensive-security / dual-use content (legitimate for authorised testing; review intended use) (test fixture / example file)## SRC028 - Giskard AI Red Teaming and LLM Security Platform
fixture
Files scanned: 80. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
- warning
body-longSKILL.md body ≈ 5686 tokens (recommended < 5000); move details to references/ - warning
missing-refreference to a missing file: scripts/check_research_mode.sh - warning
missing-refreference to a missing file: scripts/selftest/
Process rating: all ten parameters 52/100
- 0Tools and files. 2 referenced file(s) missing: scripts/check_research_mode.sh, scripts/selftest/
- 0Inputs and preconditions. Does not say what the process needs to start
- 0Failures and branches. Linear process with no failure handling
- 0Progress reporting. Says nothing while it works
- 30Running it twice. 15 mutating operations with no state check
- 70Execution cost. Instruction body is 5686 tokens
- 85Steps. 217 steps, 2 vague phrases
- 100Result and completion. Output format and completion criterion are stated
- 100When it triggers. States when to use and when not to
- 100Consistency. Name and required fields are in place
- low 12 top-level sections: this looks like several domains in one skill
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
- -42 reference files, but SKILL.md never points to them: the model will not open them
- +2Single-language instructions
- +3Description length 348: enough signal without eating the budget
- +4Structure: 35 headings
- +3Step-by-step instructions: 217 items
- +4Has examples (4 code blocks)
- +1License stated
Quality base 70; lint remarks subtract, signals add up to 100. Result: 62.