T03 · Remote Payload Retrieval and Execution
- Location
references/onboarding-utility.md:14- Finding
Unpinned Remote Onboarding Scripts Are Downloaded and Executed with Privileged Credentials
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
This is a legitimate DevOps skill, but it needs review because it can drive production changes and handles powerful GitHub, Kubernetes, and app secrets with weak safeguards.
Install only if you trust the essesseff vendor and are prepared to grant the agent/operator real DevOps authority. Pin and inspect the onboarding utility before running it, use short-lived least-privilege tokens, keep kubectl pointed only at the intended cluster, require explicit confirmation for PROD deployment, package deletion, retention changes, and repository creation, and do not store .env or notifications-secret.yaml inside Git working trees.
references/onboarding-utility.md:14Unpinned Remote Onboarding Scripts Are Downloaded and Executed with Privileged Credentials
references/onboarding-utility.md:115GitHub and Argo CD Secrets Are Written into Git Repository Working Trees Without Effective Ignore Protection
The skill explicitly states that production deployment can be triggered through the API without OTP, yet it does not pair that capability with strong warnings or approval requirements. In a high-automation agent context, this materially raises the risk of accidental or unauthorized production changes because a single API call can promote an image directly to PROD.
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
| POST | `.../apps/{app_name}/environments/QA/declare-rejected` | Mark QA image as REJECTED |
| POST | `.../apps/{app_name}/environments/STAGING/deploy-stable` | Deploy STABLE to STAGING |
| POST | `.../apps/{app_name}/environments/PROD/deploy-stable` | Deploy STABLE to PROD (no OTP) |
| GET | `.../apps/{app_name}/notifications-secret` | Get notifications-secret.yaml for Argo CD |
| GET | `.../apps/{app_name}/packages/delete-packages` | Get expired package deletion payloads |
| GET | `.../apps/{app_name}/retention-policies` | List retention policies (filter by `?state=`) |
| PATCH | `.../apps/{app_name}/retention-policies?state={state}` | Create or update a retention policy for a lifecycle state |
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
| POST | `.../apps/{app_name}/environments/QA/declare-rejected` | Mark QA image as REJECTED |
| POST | `.../apps/{app_name}/environments/STAGING/deploy-stable` | Deploy STABLE to STAGING |
| POST | `.../apps/{app_name}/environments/PROD/deploy-stable` | Deploy STABLE to PROD (no OTP) |
| GET | `.../apps/{app_name}/notifications-secret` | Get notifications-secret.yaml for Argo CD |
| GET | `.../apps/{app_name}/packages/delete-packages` | Get expired package deletion payloads |
| GET | `.../apps/{app_name}/retention-policies` | List retention policies (filter by `?state=`) |
| PATCH | `.../apps/{app_name}/retention-policies?state={state}` | Create or update a retention policy for a lifecycle state |
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
| POST | `.../apps/{app_name}/environments/QA/declare-rejected` | Mark QA image as REJECTED |
| POST | `.../apps/{app_name}/environments/STAGING/deploy-stable` | Deploy STABLE to STAGING |
| POST | `.../apps/{app_name}/environments/PROD/deploy-stable` | Deploy STABLE to PROD (no OTP) |
| GET | `.../apps/{app_name}/notifications-secret` | Get notifications-secret.yaml for Argo CD |
| GET | `.../apps/{app_name}/packages/delete-packages` | Get expired package deletion payloads |
| GET | `.../apps/{app_name}/retention-policies` | List retention policies (filter by `?state=`) |
| PATCH | `.../apps/{app_name}/retention-policies?state={state}` | Create or update a retention policy for a lifecycle state |
A documented endpoint whose purpose is to retrieve a secret-bearing YAML file is inherently credential-sensitive. In a general-purpose agent skill, exposing such an endpoint without strict guidance on confirmation, redaction, and safe handling makes credential access easier to trigger and increases the chance of accidental exfiltration.
Get notifications-secret.yaml for Argo CD setup
GET /api/v1/accounts/{account_slug}/organizations/{organization_login}/apps/{app_name}/notifications-secret
The description explicitly states that the endpoint returns Argo CD Notifications Event Processor API keys and environment-specific app secrets, and that secrets may be automatically issued during generation. That combination expands both exposure and side effects: a read-like action not only discloses credentials but may mint new ones, magnifying abuse potential.
### Description
Retrieves the notifications-secret.yaml file content for Argo CD setup. Returns YAML content with `Content-Type: application/x-yaml`. This file contains Argo CD Notifications Event Processor API keys and environment-specific app secrets.
**Note:** Secrets are automatically issued from the database when this file is generated if they don't already exist.
The example demonstrates downloading the secret file directly, which normalizes handling live credentials in a local artifact. In agent-driven environments, that pattern can cause secrets to be written to insecure disks, workspaces, or logs unless strong safeguards are built around file handling.
curl -X GET
"https://www.essesseff.com/api/v1/accounts/my-team/organizations/my-apps-org/apps/my-app/notifications-secret"
-H "X-API-Key: ess_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
-o notifications-secret.yaml
### Response
The response example includes a Kubernetes Secret manifest containing an API key and multiple app secrets in plaintext stringData. Even though the values are placeholders, this documents a workflow centered on raw credential material and is dangerous in this skill context because agents may expose, copy, or persist the manifest contents.
Returns YAML content with Content-Type: application/x-yaml:
# notifications-secret.yaml
# Store Argo CD Notifications Event Processor API key and environment-specific app secrets
# ***DELETE THIS FILE AFTER APPLYING - DO NOT COMMIT SECRETS TO GITHUB***
The documentation advertises an endpoint that retrieves notifications-secret.yaml, which is explicitly secret-bearing material for Argo CD setup. In an agent skill that can automate API access, exposing or normalizing retrieval of secret content materially increases the chance of credential disclosure, accidental logging, or misuse if the agent fetches and surfaces the YAML to users.
##### GET .../apps/{app_name}/notifications-secret
Retrieve the notifications-secret.yaml file content for Argo CD setup. Returns YAML content.
#### Other Endpoints
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
Returns GitHub API payloads for deleting package versions that correspond to expired essesseff images. The function validates that all uniquely tagged images associated with a particular GitHub package version are expired before returning the package version for deletion.
**Important:** This endpoint only returns deletion payloads. You must execute the deletion using a GitHub Personal Access Token (PAT) with appropriate permissions, as GitHub Apps cannot reliably delete package versions via the API.
### Path Parameters
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
Returns GitHub API payloads for deleting package versions that correspond to expired essesseff images. The function validates that all uniquely tagged images associated with a particular GitHub package version are expired before returning the package version for deletion.
**Important:** This endpoint only returns deletion payloads. You must execute the deletion using a GitHub Personal Access Token (PAT) with appropriate permissions, as GitHub Apps cannot reliably delete package versions via the API.
### Path Parameters
The skill documents a workflow that returns fully formed deletion payloads for package versions and then instructs callers to execute those DELETE operations against GitHub. In an agent context, this increases the risk of destructive action execution from tool-derived parameters without sufficient human confirmation, scope validation, or allowlisting, potentially causing irreversible package deletion in external repositories.
"organization_login": "my-apps-org",
"repository_name": "my-app",
"package_name": "my-app",
"github_api_endpoint": "DELETE /repos/{owner}/{repo}/packages/{package_type}/{package_name}/versions/{package_version_id}",
"github_api_params": {
"owner": "my-apps-org",
"repo": "my-app",
The documentation explicitly instructs users to copy notifications-secret.yaml into a cloned Git repo during Argo CD setup. Storing or staging secret manifests in repository working trees materially increases the risk of accidental commit, push, CI exposure, or local disclosure of cluster credentials and notification secrets.
### Argo CD Setup
For each environment:
1. (Subscribers) Downloads `notifications-secret.yaml`.
2. Clones the argocd-env repo (`{app-name}-argocd-{env}`).
3. Creates `.env` file with required variables.
4. (Subscribers) Copies `notifications-secret.yaml` to repo.
The utility writes sensitive values such as GITHUB_TOKEN and Argo CD machine-user credentials into .env files inside cloned repos before running setup-argocd.sh. Even if intended for local automation, placing credentials in repo directories creates a high chance of accidental commit, leakage through shell history, editor indexing, backups, or CI/CD tooling.
1. (Subscribers) Downloads `notifications-secret.yaml`.
2. Clones the argocd-env repo (`{app-name}-argocd-{env}`).
3. Creates `.env` file with required variables.
4. (Subscribers) Copies `notifications-secret.yaml` to repo.
5. Runs `setup-argocd.sh`.
**Variables written to `.env` files:**
Tool calls are chained to bypass individual safety checks or escalate capabilities beyond what any single tool call would allow.
Install on Ubuntu/Debian:
sudo apt-get update && sudo apt-get install -y curl git jq
# kubectl: https://kubernetes.io/docs/tasks/tools/install-kubectl-linux/
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
- Any K8s cluster works, including a single-VM `k3s` setup.
- Argo CD must be installed. The `setup-argocd-cluster.sh` script in this repo provides an easy way to do this.
- `kubectl` must be configured (`~/.kube/config` or `KUBECONFIG` env var) and pointing at the correct cluster for each environment **before** running `--setup-argocd`.
- The destination K8s namespace defaults to the `GITHUB_ORG` string. If your org name isn't DNS compliant or is 63+ characters, set `K8S_NAMESPACE` in `.essesseff` to a valid override.
## essesseff API Key (subscribers only)
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
- Any K8s cluster works, including a single-VM `k3s` setup.
- Argo CD must be installed. The `setup-argocd-cluster.sh` script in this repo provides an easy way to do this.
- `kubectl` must be configured (`~/.kube/config` or `KUBECONFIG` env var) and pointing at the correct cluster for each environment **before** running `--setup-argocd`.
- The destination K8s namespace defaults to the `GITHUB_ORG` string. If your org name isn't DNS compliant or is 63+ characters, set `K8S_NAMESPACE` in `.essesseff` to a valid override.
## essesseff API Key (subscribers only)
The README advertises automation for creating repositories, creating apps, and configuring Argo CD across environments without an adjacent, explicit warning that these actions change remote infrastructure and may create or modify production-facing resources. In a DevOps skill, omission of impact warnings increases the chance of accidental high-impact operations by users or downstream agents.
npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
BASE=https://www.essesseff.com/api/v1
# List global templates
curl -H "X-API-Key: $ESSESSEFF_API_KEY" "$BASE/global/templates"
# Create an app
curl -X POST "$BASE/accounts/$ACCOUNT/organizations/$ORG/apps?app_name=my-app" \
The README includes a concrete command to deploy a STABLE image to PROD but does not place a clear production-impact warning immediately around the example. In a skill that may be followed by an autonomous agent or rushed operator, this materially increases the risk of accidental production deployment, service disruption, or unapproved release activity.
The skill documents an endpoint that returns package deletion payloads but provides no user-facing warning that acting on those payloads can remove artifacts from external registries. In an agent setting, omission of destructive-impact warnings increases the chance of unintended deletion when the model interprets the capability as routine maintenance.
The onboarding utility can create repositories, create apps, and configure Argo CD across environments, all of which modify external systems, but the documentation does not emphasize that these actions are state-changing and potentially irreversible. For an autonomous or semi-autonomous agent, that lack of warning makes unintended infrastructure changes more likely.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -X GET \
"https://www.essesseff.com/api/v1/accounts/my-team/organizations/my-apps-org/apps" \
-H "X-API-Key: ess_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
The app-creation endpoint is documented as creating 9 repositories, setting webhooks, assigning teams, and configuring retention policies, but the docs do not emphasize that this is a high-impact side-effecting operation requiring explicit user confirmation. In an agent skill context, this omission increases the risk that an automated workflow could trigger broad infrastructure changes from an ambiguous or incomplete user request.