Decepticon is an autonomous red-team agent that coordinates AI agents, security tools, sandboxes, and supporting services for authorized cybersecurity assessments. Security researchers and red teams can run it through its Docker stack, cloud service, command-line interface, or Python SDK, with the catalogue entries representing its available skills.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/purpleailab/decepticon/final-reportnpx skills add PurpleAILAB/Decepticon --skill final-reportgit clone --depth 1 https://github.com/PurpleAILAB/DecepticonWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/purpleailab/decepticon/final-report)<a href="https://agentmods.dev/skills/purpleailab/decepticon/final-report"><img src="https://agentmods.dev/badge/skills/purpleailab/decepticon/final-report.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00028 | $0.04836 |
| Opus 5 | $0.00014 | $0.02418 |
| Sonnet 5 | $0.00006 | $0.00967 |
| Haiku 4.5 | $0.00003 | $0.00484 |
Grade A, and why
final-report scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 518 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Final Engagement Report Generation
Execute when all OPPLAN objectives are in passed or blocked status. Output is two Markdown documents in report/.
Report Generation Workflow
1. Read all findings/*.md (parse YAML frontmatter for metadata)
2. Read all findings/attack-paths/PATH-*.md
3. Read timeline.jsonl (activity log)
4. Read plan/roe.json + plan/conops.json + plan/deconfliction.json (engagement context)
5. Read plan/opplan.json (objective status summary)
6. Generate report/executive-summary.md
7. Generate report/technical-report.md
8. Do not create a legacy findings.md summary; summaries belong in report/
Step 1: Initialize Output Directory
bash(command="mkdir -p /workspace/report", description="Create the report output directory")
Step 2: Aggregate Findings
Read every findings/FIND-*.md (canonical name findings/FIND-{NNN}.md; the
id frontmatter field matches the filename) and extract frontmatter fields:
for each FIND-NNN.md:
- id, title, severity, cwe, cvss_score, cvss_vector, vrt,
target (affected_target), affected_component, phase,
technique (ATT&CK ID), status, confidence
Sorting rules:
- Primary: severity (
critical→high→medium→low→informational) - Secondary: CVSS score descending
- De-duplicate findings that appear across multiple objectives (keep highest-severity instance)
- Flag
confidence: unverifiedfindings — they require a disclaimer note in the report
Step 3: Generate Executive Summary
Write to report/executive-summary.md using the template below. Use plain business language throughout — zero technical jargon.
# Executive Summary
## [Engagement Name] — Penetration Test Results
**Prepared for**: [Client Name]
**Engagement Period**: [Start Date] – [End Date]
**Conducted by**: Decepticon Autonomous Red Team
**Classification**: CONFIDENTIAL
---
## Engagement Overview
[2–3 sentences: what was tested, why, what methodology was used.
Example: "Between [dates], [client] engaged the red team to assess the
security posture of [scope]. The assessment simulated a [threat actor
archetype from conops.json] targeting [primary objective from roe.json].
Testing followed the Penetration Testing Execution Standard (PTES) and
MITRE ATT&CK framework."]
---
## Key Metrics
| Metric | Value |
|--------|-------|
| Total Findings | [N] |
| Critical | [N] |
| High | [N] |
| Medium | [N] |
| Low / Informational | [N] |
| Objectives Completed | [N] / [total] |
| Objectives Blocked | [N] / [total] |
| Attack Success Rate | [%] |
| Detection Rate | [%] (findings detected by existing controls) |
---
## Top 3 Critical Findings
### 1. [FIND-ID]: [Title]
[1 paragraph, business impact focus. No CVE numbers, no tool names.
Example: "Attackers with internet access can gain full administrative
control of the company's customer database without any valid credentials.
This would allow theft of all customer records, potential regulatory
fines under GDPR, and complete loss of customer trust."]
### 2. [FIND-ID]: [Title]
[1 paragraph, business impact focus]
### 3. [FIND-ID]: [Title]
[1 paragraph, business impact focus]
*[If fewer than 3 critical findings, use highest-severity findings available.]*
---
## Overall Risk Rating
**[CRITICAL / HIGH / MEDIUM / LOW]**
[2–3 sentences justifying the rating in business terms. Consider: ease of
exploitation, potential business impact, data sensitivity, regulatory exposure.
Example: "The overall risk is rated CRITICAL because an unauthenticated
attacker on the internet achieved domain administrator access within [X]
minutes, with no detection by existing security controls. This represents
a complete failure of the network perimeter and would result in total
compromise of all company systems."]
---
## Strategic Recommendations
1. **[Recommendation 1 — address most critical finding]** — [1 sentence, business language, e.g. "Immediately restrict internet access to the database server and require multi-factor authentication for all administrative accounts."]
2. **[Recommendation 2]** — [1 sentence]
3. **[Recommendation 3 — detection/monitoring gap]** — [1 sentence]
4. **[Recommendation 4 — process/architecture]** — [1 sentence]
5. **[Recommendation 5 — longer-term posture improvement]** — [1 sentence]
---
*Full technical details, evidence, and step-by-step remediation instructions are provided in the Technical Report.*
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 518 lines · 28 tokens per session scan A ce07fac6096a
final-report is a skill published in the GitHub repository PurpleAILAB/Decepticon (5,450 stars, last pushed 6d ago), licensed Apache-2.0. It adds 28 tokens to every session and 4,836 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
interactive-dashboard
Interactive web dashboards: stock trackers, sector heatmaps, portfolio monitors — served via preview URL.
onboarding
First-time user onboarding to set up investment profile, watchlists, portfolio, and preferences.
idea-generation
Stock screening and idea generation: quantitative screens, thematic analysis, shortlist.
secretary
Workspace and research management — dispatch analyses, monitor running agents, manage workspaces and threads.
add-model
Add a new language model to the Giselle codebase. Use when the user wants to add, register, or integrate a new LLM model (OpenAI, Anthropic, Google) into the system.
python-lib-analyzer
Analyze any Python library structure, explore modules, classes, and functions with signatures and documentation.