Agent Economy ReportAgent Economy ReportCAUTION version 1.0.1 · audited 2026-09-17
This skill is a collection of markdown instructions that teach an agent to be more proactive by anticipating next steps, recovering from interruptions, and keeping notes in a local ~/proactivity folder. It does not contain executable code, network calls, credential capture, or payment handling. Setup instructions use shell commands to create files and set folder permissions, and they propose adding behavior-changing text to AGENTS, SOUL, TOOLS, and HEARTBEAT files only after showing the exact text and getting approval. The package repeatedly says it will not send messages, spend money, delete data, or modify files outside ~/proactivity without approval. The behavior matches its declared purpose, but the workspace integration should be reviewed because it can change how the agent behaves.
| Declared purpose | Anticipates needs, keeps work moving, and improves through use so the agent gets more proactive over time. |
| Observed behavior | Markdown guidance only; instructs the agent to create and maintain local state under ~/proactivity, read that state during recovery, follow DO/SUGGEST/ASK/NEVER boundaries, and optionally propose edits to AGENTS, TOOLS, SOUL, and HEARTBEAT after explicit approval. No runtime network or executable payloads were found. |
| Verdict, rules only | SAFE |
| Verdict, AI | CAUTION |
| Final verdict | CAUTION (never better than either pass) |
| capability | vs purpose | detail |
|---|---|---|
| filesystem_write | needed | Creates and maintains local files under ~/proactivity using mkdir/touch in setup.md lines 106-113 and memory-template.md lines 55-59; with approval it may write behavior snippets to AGENTS/TOOLS/SOUL/HEARTBEAT. |
| filesystem_read | needed | Reads ~/proactivity memory, session state, and working buffer during recovery, as described in recovery.md lines 7-9, and may read behavior files AGENTS, TOOLS, SOUL, and HEARTBEAT if the user wants integration (SKILL.md line 121). |
| shell | needed | Setup and migration guides provide bash commands using mkdir, touch, and chmod to initialize local state and set permissions (memory-template.md lines 55-59; setup.md lines 106-113; migration.md lines 29-33). |
| persistence | needed | Durable local state is stored in ~/proactivity and reused across sessions (SKILL.md lines 13 and 95), though this is ordinary file storage rather than operating-system persistence or autostart. |
| network | exceeds purpose | No network capability is used; the only domains are passive profile/homepage links at SKILL.md line 5 and skill-card.md line 9. |
| severity | finding |
|---|---|
| info | Declared purpose This matches the package's stated purpose and the description on the skill card. SKILL.md:6 description: Anticipates needs, keeps work moving, and improves through use so the agent gets more proactive over time. |
| info | No network requirement The skill explicitly says it has no network requirement. The only URLs found are passive links, not network calls made by the skill. SKILL.md:119 It does not require network access by itself. |
| info | High-impact actions require approval The skill states it will not spend money, send messages, delete data, or make commitments without approval. No wallet or payment code was found. SKILL.md:102 sends messages, spends money, deletes data, or makes commitments without approval |
| low | SOUL edit requires approval Before adding the SOUL steering snippet, the skill instructs the agent to show the exact text and wait for explicit approval. setup.md:13 Show the exact snippet immediately and wait for explicit approval before writing it. |
| medium | Proposes modifying AGENTS behavior file This snippet would add proactivity routing to the workspace AGENTS file. It is outside ~/proactivity and could change agent behavior, so a user should review it carefully before approval. setup.md:34 - **Proactivity:** `~/proactivity/` (via `proactivity` skill) - proactive operating state, action boundaries, active task recovery, and follow-through rules |
| low | TOOLS edit is not automatic The setup guide explicitly tells the agent not to edit the TOOLS file automatically, and to show the snippet and wait for approval. setup.md:88 Do not edit the workspace TOOLS file automatically. |
| low | Boundary for external and irreversible actions This guardrail covers spending, deletion, scheduling, and contacting people. It is consistent with the claimed purpose and reduces side-effect risk. setup.md:98 For send, spend, delete, reschedule, or contact actions, stop and ask first |
| low | Shell setup creates local files and permissions The setup uses shell commands to create local state directories and set restrictive permissions. The commands stay under the ~/proactivity path and do not install software or change system startup. setup.md:110 chmod 700 ~/proactivity ~/proactivity/domains ~/proactivity/memory |
| info | Local state initialization command The memory template includes a bash command to create the proactivity directory structure. It does not reach outside the home directory. memory-template.md:56 mkdir -p ~/proactivity/{domains,memory} |
| info | Recovery reads local state The recovery flow instructs the agent to reconstruct recent work from local proactivity state before asking the user to repeat anything. recovery.md:7 1. Read stable proactivity memory |
| low | Publisher acknowledges behavior-file risk This known-risk note matches the main caution for this package: approving those proposed edits changes agent behavior. skill-card.md:29 Workspace behavior files such as AGENTS, TOOLS, SOUL, or HEARTBEAT can alter how the agent behaves if edited without careful review. |
| info | Publisher profile link This is a passive profile URL. It is not fetched or contacted by any skill code. skill-card.md:9 [ivangdavila](https://clawhub.ai/user/ivangdavila) |
0 AI finding(s) were dropped because their file, line or quote did not match the package.
| domain | service rating | where |
|---|---|---|
| clawhub.ai | not in the service index | skill-card.md:9 |
| clawic.com | not in the service index | SKILL.md:5 |
| Binaries invoked | bash, chmod |
| Environment variables read | none |
| Hard-coded wallets | none |
| Pipes a download to a shell | 0 |
| eval / exec / subprocess | 0 |
| base64 blobs | 0 |
| File writes | 0 |
| Persistence | 0 |
| file | lines | sha256 |
|---|---|---|
SKILL.md | 136 | 2ebb2ca0988db725… |
_meta.json | 6 | 4cab4528441cf5f4… |
boundaries.md | 42 | b46480181f597f9a… |
execution.md | 86 | 67984a39fa3fea99… |
heartbeat-rules.md | 37 | ed9a0ef558086fa2… |
memory-template.md | 78 | 6a1e1eb4a15d7cad… |
migration.md | 41 | f4b750456e66b169… |
recovery.md | 28 | 876c61a40744b3ed… |
setup.md | 135 | 6ddf5236923fa550… |
signals.md | 42 | fc70d2430c9138da… |
skill-card.md | 67 | 877660e6b07ae4b4… |
state.md | 36 | 685c33dbb42ad6eb… |
For agents
JSON: https://agenteconomy.report/k/proactivity.audit.json · badge: https://agenteconomy.report/k/proactivity.audited.svg ·
skill rating: /k/proactivity · commission an audit of any skill: US$ 29 per version.
The complete published package of this exact version was downloaded from the registry and read statically; nothing was executed. A deterministic pass extracts network destinations, binaries, environment variables, writes, install commands, obfuscation markers, subprocesses and wallets, each with file and line. An AI then reads the whole package with those facts and writes the summary, the capabilities and the findings under a strict schema; every finding it produces must cite a file, a line and the exact text, or it is dropped. The final verdict is the worst of the two passes. The audit does not change the skill's trust tier (policy); the author may respond through the dispute channel and the response is published here. Commissioned by: the Agent Economy Report (free program: the 150 most downloaded skills, September 2026).