Soul HaulA field guide to agent souls

The files people write to give AI agents a personality, collected from the public web every day.

The daily 5Submit a soul
  • Index911
  • New this week
  • Saved

Kinds

  • Soul879
  • Agent rules9
  • Persona18
  • Bot profile11
  • Framework config3

Archetypes

all
  • Blunt expert148
  • Friendly helper44
  • Coding pair109
  • Research analyst116
  • Creative writer7
  • Companion / character101
  • Personal clone3
  • Business operator336
  • Teacher / coach20
  • Experimental / weird26

Agents

  • OCOpenClaw762
  • HEHermes743
  • BOBrowserOS679
  • OTOther or custom110
  • CCClaude Code11
  • CXCodex5
  • CLClaude3
  • GKGrok2
  • CRCrewAI1
  • CUCursor1
AboutPrivacy
Soul HaulA field guide to agent souls

Soul Haul catalogs agent personality files from public sources. Every entry links back to its original, and authors can have theirs removed at any time.

AboutArchetypesNewsletterTakedownPrivacy
  • Index911
  • New this week
  • Saved

Kinds

  • Soul879
  • Agent rules9
  • Persona18
  • Bot profile11
  • Framework config3

Archetypes

all
  • Blunt expert148
  • Friendly helper44
  • Coding pair109
  • Research analyst116
  • Creative writer7
  • Companion / character101
  • Personal clone3
  • Business operator336
  • Teacher / coach20
  • Experimental / weird26

Agents

  • OCOpenClaw762
  • HEHermes743
  • BOBrowserOS679
  • OTOther or custom110
  • CCClaude Code11
  • CXCodex5
  • CLClaude3
  • GKGrok2
  • CRCrewAI1
  • CUCursor1

HE

Hermes

KindAllSoul 738Persona 2Agent rules 1Bot profile 1Framework config 1
ArchetypeAllBusiness operator 263Blunt expert 141Research analyst 102Coding pair 78Companion / character 74Friendly helper 39Experimental / weird 21Teacher / coach 15Creative writer 7Personal clone 3

743 souls, by quality score.

181 · № F755·41E7 · Friendly helper

★ 41

Patient starter guide for learning Hermes

Sets up a patient, practical guide that explains terms in plain language, recommends one default approach, verifies tool results before claiming success, and stays within a small disabled-capability toolset.

patientpracticalwarmHESoulGitHubOct 21, 2026
View

182 · № 8645·D892 · Companion / character

★ 19

Efficient bilingual Go assistant MalikClaw for edge hardware

Sets up MalikClaw, a lean, action-oriented assistant that acts with tools first, keeps answers concise and factual, protects user privacy, and switches to respectful Urdu when requested.

warmconcisedirectOCHEBOSoulGitHubOct 22, 2026
View

183 · № C3B4·1A49 · Blunt expert

★ 4

Direct, act-first assistant persona for OpenClaw

Sets up an opinionated assistant that skips filler, acts on information requests without clarifying questions, and treats access to a user's private data with respect. It asks before sending anything public on the user's behalf.

directopinionatedcandidOCHEBOSoulGitHubOct 21, 2026
View

184 · № CB41·A5E3 · Companion / character

★ 8

Continuous shared assistant soul across runtimes

Sets up one persistent assistant identity that stays the same across models and runtimes, with preferences, judgment, privacy-first habits and a close but not overstepping relationship with the user. Memory and source attribution track what happened and who recorded it.

warmdirectmeasuredOCHEBOSoulGitHubOct 22, 2026
View

185 · № 2512·BF53 · Coding pair

★ 59

Pragmatic SRE collaborator that answers conclusion-first in Chinese

Sets up a reliable, pragmatic SRE partner that gives the conclusion first, supporting evidence and next steps, avoids filler, flags uncertainty with verification paths, and stays cautious with high-risk changes.

directprofessionalpragmaticOCHEBOSoulGitHubOct 22, 2026
View

186 · № E8DB·3EE7 · Blunt expert

★ 388

Direct, opinionated SOUL.md persona for OpenClaw agents

Sets up a direct, opinionated working style with no flattery or filler that matches the user's energy. It also includes memory behavior rules (search before claiming forgetfulness, never store credentials) and anti-patterns like not asking permission for low-risk actions.

directbluntcasualOCHEBOSoulGitHubOct 22, 2026
View

187 · № B109·7554 · Research analyst

★ 315

Precise legal assistant with plain-language explanations

Sets up Lavern as a careful legal-document assistant that explains complex legal concepts in plain language first, cites specific text, admits uncertainty, and stays conservative on risk while formal in deliverables. It is written for CLI or Claw mode with a browser profile taking priority when available.

preciseapproachableformalOCHEBOSoulGitHubOct 22, 2026
View

188 · № D2BB·5AB0 · Business operator

★ 726

Warm customer success lead Sam for an agent team

Sets up Sam, a patient customer success and onboarding lead who triages tickets, separates bugs from confusion, and escalates recurring issues with evidence. Tone is warm, concrete and non-condescending, with a clear emphasis on follow-through.

warmpatientempatheticOCHEBOSoulGitHubOct 21, 2026
View

189 · № 8D8A·7415 · Friendly helper

★ 726

Dara, product designer persona for OpenClaw

Sets up Dara, a curious and warm product designer and UX researcher who starts from user problems, values consistency and craft, and explains trade-offs with concrete examples. The file is adapted from a community agent template with light edits.

warmcuriousspecificOCHEBOSoulGitHubOct 21, 2026
View

190 · № 4953·5179 · Companion / character

★ 494

Calm, predictable companion for a Unitree Go2-W robot

Sets up a calm, capable mobile companion robot that announces its movements, moves slowly and predictably around people, stays curious without intruding, and speaks briefly.

calmbriefpredictableOCHEBOSoulGitHubOct 21, 2026
View

191 · № A89B·3225 · Research analyst

★ 3.0k

Source-verifying regulatory fact-checker persona

Sets up an agent that checks factual claims against authoritative regulatory sources, verifies rule numbers, effective dates and citations, and explicitly flags anything it cannot verify rather than assuming it is correct.

preciseskepticalcarefulOCHEBOSoulGitHubOct 21, 2026
View

192 · № 3357·CD31 · Business operator

★ 294

Direct operator assistant for reviewable chat and issue workflows

Sets up a concise, action-oriented assistant that turns operator requests into bounded, reviewable steps, keeps evidence and blockers visible, routes work when another agent fits better, and preserves lessons from accepted work.

directconciseplainOCHEBOSoulGitHubOct 21, 2026
View

193 · № E86C·213A · Research analyst

★ 11

Vigilant phishing detection analyst for email and URL triage

Sets up a cautious security analyst that scores emails and URLs for phishing indicators using structure-only analysis, flags uncertain cases for human review, explains its red flags to users, and produces periodic phishing trend reports.

vigilantcautiouspreciseOCHEBOSoulGitHubOct 21, 2026
View

194 · № 7E7D·4D86 · Business operator

★ 3

Methodical reliability-first infrastructure agent

Sets up an infrastructure agent that treats reliability as the product: it designs for failure, insists on version-controlled, immutable, least-privilege systems, and raises a fixed list of red flags such as committed secrets, single points of failure and missing health checks. It is methodical, proactive and honest about trade-offs.

methodicalproactivehonestOCHEBOSoulGitHubOct 22, 2026
View

195 · № D866·790D · Blunt expert

seen in 2 places

Opinionated, resourceful SOUL.md persona for OpenClaw with terse Telegram mode

Sets up an opinionated, resourceful assistant that skips filler, acts boldly on internal tasks, and stays careful with external actions and private data. It also defines a terse progress-bar style for a Telegram chat and a no-em-dash writing rule.

directterseplayfulOCHEBOSoulGitHubOct 21, 2026
View

196 · № F6CE·5E64 · Blunt expert

seen in 2 places

Resourceful, opinionated SOUL.md assistant for OpenClaw

Sets up a resourceful, opinionated assistant that solves obstacles before reporting them, skips filler and flattery, and treats access to a user's private data with care. It asks before external actions and keeps group-chat replies cautious.

directcandidresourcefulOCHEBOSoulGitHubOct 21, 2026
View

197 · № 9E62·7F98 · Research analyst

seen in 2 places

Post-task review engine that analyzes logs offline and never interrupts

Sets up a calm, evidence-based retrospective agent that runs only after a task ends, reviews full logs against the blueprint, gives constructive criticism of work rather than of agents, and turns lessons into updates to AGENTS.md and MEMORY.md.

objectivecalmconstructiveOCHEBOSoulGitHubOct 21, 2026
View

198 · № FD0E·D9D1 · Business operator

seen in 2 places

LinkedIn content creator focused on B2B thought leadership

Sets up an opinionated LinkedIn content creator that drafts posts in the user's voice around one narrow reader and one takeaway, avoids clichés, flags stats for verification, and saves drafts and retros to workspace files. It also requires approval before posting and does not invent anecdotes.

opinionateddirectgroundedOCHEBOSoulGitHubOct 21, 2026
View

199 · № EA5C·18E2 · Blunt expert

★ 48

Resourceful, opinionated personal assistant with guest-level privacy respect

Sets up a direct, competent assistant that skips filler, forms opinions, tries to solve problems before asking, and treats access to a user's private data as a trust to protect. External actions get more caution than internal ones.

directconciseopinionatedOCHEBOSoulGitHubOct 22, 2026
View

200 · № 807C·0FBB · Coding pair

★ 48

Disciplined software engineer persona for SOUL.md agents

Sets up a careful, direct software engineer who values correctness, understands problems before coding, asks when unclear, and keeps scope tight. Written as a SOUL.md identity file with style, avoidances and technical preferences.

directprecisedisciplinedOCHEBOSoulGitHubOct 22, 2026
View

201 · № 2D35·5420 · Business operator

★ 1

SRE runbook writer for sleep-deprived on-call engineers

Sets up a seasoned SRE persona that writes precise, numbered incident runbooks assuming nothing about the reader's context. Every runbook needs a fixed set of sections, exact CLI commands with marked placeholders, verification checks and time estimates.

precisemethodicalclearOCHEBOSoulGitHubOct 21, 2026
View

202 · № 08E9·079F · Coding pair

★ 8

Finding triager that deduplicates and scores security findings

Sets up a triage agent that removes noise from bulk security findings, merges duplicates and related issues on the same locus, and assigns defensible CVSS scores and severity bands. It is a strictly procedural working style with no conversational voice.

terseformalpreciseOCHEBOSoulGitHubOct 21, 2026
View

203 · № A79D·A622 · Blunt expert

★ 41

Competent, opinionated assistant with ClawNet identity boundaries

Sets up a direct, opinionated assistant that skips filler, acts boldly on internal tasks but carefully on external ones, and keeps one user's social identity and domain information separate from other agents and domains.

directopinionatedconciseOCHEBOSoulGitHubOct 21, 2026
View

204 · № 2A5E·B4D4 · Teacher / coach

★ 7

Patient teacher persona for studying and quizzes

Sets up a patient tutor that explains from intuition, checks understanding, quizzes with targeted feedback, and builds study plans around the learner's goals and time. Encouraging without flattery.

patientencouragingcuriousOCHEBOSoulGitHubOct 22, 2026
View

205 · № 145C·2176 · Research analyst

★ 80

Structure-first realist analyst persona

Sets up an analytical, engineering-minded voice that breaks problems into structure, weighs incentives, costs and execution conditions, and explains complex issues clearly. It allows light irony and judgment but avoids performative answers.

analyticalpragmaticdryOCHEBOSoulGitHubOct 22, 2026
View

206 · № A5D1·ACA2 · Business operator

★ 123

Skeptical sales-deal verifier that checks claims against data

Sets up a skeptical, numbers-driven agent that tests sales reps' optimistic claims against signed documents and customer behavior, and reconciles conflicting sources before reaching a conclusion.

skepticalanalyticaldirectOCHEBOSoulGitHubOct 21, 2026
View

207 · № 7C29·876A · Business operator

★ 36

Internal customer-service assistant that supports agents without replying to customers

Sets up an internal assistant that helps customer service staff look up scripts and business knowledge, answers only from configured sources, never fabricates content, and never replies to customers directly. It suggests workflows and escalation but leaves decisions to staff.

professionalreliablepracticalOCHEBOSoulGitHubOct 21, 2026
View

208 · № 5341·DC20 · Business operator

★ 36

Reliable publisher for Toutiao drafts (SOUL.md agent)

Sets up an execution-focused agent that publishes only human-approved drafts to Toutiao, logs every result for analytics, never stores credentials in the workspace, and respects platform rules and rate limits.

reliabletersemethodicalOCHEBOSoulGitHubOct 21, 2026
View

209 · № 4EE6·D235 · Research analyst

★ 36

Toutiao hot-topic monitor that reports structured daily digests

A data-driven monitoring persona for Toutiao that searches hot topics and viral articles and produces stable, structured daily reports for downstream tools. It never publishes or operates accounts, and it answers concisely with sources and time ranges noted.

cleardata-driventerseOCHEBOSoulGitHubOct 21, 2026
View

210 · № ADB1·B7CB · Business operator

★ 36

Scope-bound publisher agent for Baijiahao drafts

A terse, operational agent that publishes approved drafts to Baijiahao on the user's behalf and reports success or failure with next steps. It refuses monitoring, viral breakdown and rewriting, and does not publish without user consent.

terseoperationalconciseOCHEBOSoulGitHubOct 21, 2026
View
← PreviousPage 7 of 25Next →